100% found this document useful (2 votes)

225 views

Sentimental Analysis

This file contains product reviews using sentimental analysis. Along with literature reviews and code used to get the results.

Uploaded by

Ashu Tiwari

Available Formats

Download as PDF, TXT or read online on Scribd

100% found this document useful (2 votes)

225 views

Sentimental Analysis

This file contains product reviews using sentimental analysis. Along with literature reviews and code used to get the results.

Uploaded by

Ashu Tiwari

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 171

Iphone Product Review Using SentimentAnalysis

Abstract

Anirudh Shankar ,Ayush Agarwal,Prakhar Chawla,Harshit Mehta,Paras Mukhija,Gulshan

Kumar,Kushagra Tiwari,Aashray Nagrath,Dr. Senthil Kumaran S

In this modern age of online shopping, every platform that offers to galvanise their
customers with exciting offers sometimes end up delivering the prodigalproduct.
Different set of customer reviews are available on the website of these online
shopping portals, but they don’t give relevant way of analysing in terms of statiscs. This
project will give its contribution in giving a numeric percentage based on the statistics
generated by the help of the customer review on a specific product. The customer on
websites of Amazon, Flipkart, Snap deal and many more, post their thoughts on the
product delivery and those data would be the key to the method. The method of
approach that will be used to generate, the base of benchmark criteria of a products
relevancy is sentimental anlaysis. The opinions in terms of text will be computationally
identified in order to determine whether the writer’s attitude towards a particular
product is positive, negative or neutral. The main objective of the project is to give a
clear path of fundaments to work on while choosing a product. The methodolgy and
idea can be used in other fields of work aswell.
Keywords:
Sentiment Analysis; Machine Learning Algorithm; Data Mining; fine Grained
Sentiments; Blogs; Contextual Polarity; Product Review
BATCHDETAILS
Serial Number Name REG NO
1) Kushagra Tiwari 16BPI0030
2) Kartik Mathur 16BME0767
3) Aashray Nagrath 16BPI0047
4) Anirudh Shankar 16BME0341
5) Prakhar Chawla 16BME2031
6) Harshit Mehta 16BME0756
7) Paras Mukhija 16BME0930
8) Gulshan Kumar 16BME0598
9) Ayush Agarwal 16BME2024
Sentiment Analysis
on I-phone
product
Presented by- Group 2
Introduction
• In this modern age of online shopping, everyplatform
that offers to galvanise their customers with exciting
offers sometimes end up delivering the prodigal
product.
• Different set of customer reviews are available onthe
website of these online shopping portals, but they
don’t give relevant way of analysing in terms of
statiscs.
• This project will give it’s contribution in giving a
numeric percentage based on the statisticsgenerated
by the help of the customer review on a specific
product.
Introductio
n
• The method of approach that will be used to
generate, the base of benchmark criteria of a
products relevancy is sentimental anlaysis.
• The opinions in terms of text will be
computationally identified in order to
determine whether the writer’s attitude
towards a particular product is positive,
negative or neutral
LITERATUREREVIEW
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes

1 Twitter Stanford The purpose of their project Use of machine learning Their hypothesis was that they
Sentiment 2009 was to build an algorithm that algorithms to classify could obtain high accuracy on
Analysis can accurately classify Twitter user sentiments on classifying sentiment in Twitter
messages as positive or twitter to research public messages using machine
negative, with respect to a opinion about company learning techniques.
query term. or a product. Generally, this type of sentiment
analysis is useful for consumers who
are trying to research
a product or service, or marketers
researching public opinion of their
company.
2 Topic School of In this paper, instead of Our investigation This task aims to automatically
Sentiment EECS, Peking presenting the sentiment illustrates that three generate the overall sentiment
Analysis in University, polarity of each tweet relevant types of informa-tion is polarity for a given hashtag in a
Twitter: A Beijing, to the topic,we focus our study useful to address the certain time period, which markedly
Graph-based on hashtag-level sentiment task, including differs from the conventional
Hashtag classification. (1)sentiment polarity of sentence-level and document-level
Sentiment tweets containing the sentiment
Classification hashtag analysis.
Approach (2)hashtags co-
occurrence relation-
ship
(3)the literal meaning of
hashtags.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes

3 A surveyon University of This paper presents a rigorous This survey covering Several sub-tasks need to be
opinion Hyderabad, survey on sentiment analysis, published literature performed for sentiment analysis
mining and Hyderabad which portrays views presented during 2002-2015, is which in turn can be accomplished
sentiment by over one hundred articles organized on the basis of using various approaches and
analysis: published in the last decade sub-tasks to be techniques.
Tasks, regarding necessary tasks, performed, machine
approaches approaches, and applications of learning and natural
and sentiment analysis. language processing
applications techniques used and
applications of sentiment
analysis. The paper also
presents open issues and
along with a summary
table of a hundred and
sixty one articles.

4 A study of University of In this paper, we explore We show that variants of Use of weighted scheme from
Information Wolverhampto whether more sophisticated the classic tf.idf scheme information retrieval to enhance
Retrieval n feature weighting schemes from adapted to sentiment accuracy of sentimental analysis.
weighting Information Retrieval can analysis provide
schemes for enhance classification accuracy. significant
sentiment increases in accuracy,
analysis especially when using
a sublinear function for
term frequency weights
and document frequency
smoothing
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes

5 A System for University of This paper describes a system sentiment analysis can It offers the public, the
Real-time Southern for real-time help explore how these media, politicians and scholars a
Twitter California analysis of public sentiment events affect public new and
Sentiment toward opinion. While timely perspective on the dynamics
Analysis of presidential candidates in the traditional content of the
2012 U.S. 2012 U.S. analysis electoral process and public opinion.
Presidential election as expressed on takes days or weeks to
Election Twitter, a microblogging complete, the
Cycle service. system demonstrated
here analyzes
sentiment in the entire
Twitter traffic about
the election, delivering
results instantly and
continuously.
6 Sentiment National This paper introduces an experiments Use of support vector machines to
analysis using Institute of approach to sentiment using a feature set bring together diverse sources of
support vector Informatics analysis which uses support enriched with topic potentially pertinent information.
machines with vector machines information
diverse (SVMs) to bring together on a smaller dataset of
information diverse sources of potentially music reviews hand
sources pertinent information, including annotated for topic are
several favourability also reported, the results
measures for phrases and of which suggest that
adjectives and, incorporating topic
where available, knowledge of information into such
the topic of the models may also yield
text. improvement
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
7 Sentiment Proceedings of User reviews and comments on They presented a web In this paper, only German reviews
Analysis for the hotels on the web are an based opinion mining from German sites are handled. So
Hotel computational important information source in system for hotel reviews the analysis systems first does a
Reviews linguistics- travel planning. Therefore, and user comments that language check (LangID) to filter
applications knowing about these comments supports the hotel out reviews in other languages than
conference. is important for quality control management in German because even
jachranka, 2011 to the hotel management, too. monitoring what is
published on the web
about their houses. The
system is capable of
detecting and retrieving
reviews on the web, to
classify and analyze
them, as well as to
generate comprehensive
overviews of these
comments.
8 Sentiment ACM New In this work they have used Their results show that Emotions have a high impact in
analysis of York, NY, lexical sentiment analysis to projects developed in productivity, task quality, creativity,
commit USA,2014 study emotions expressed in Java tend to have more group rapport and job satisfaction.
comments in commit comments of different negative commit They have used lexical sentiment
GitHub: an open source projects and analyze comments, and that analysis to study emotions expressed
empirical their relationship with different projects that have more
study factors such as used distributed teams tend
programming language, time to have a higher
and day of the week in which positive polarity in their
the commit was made, team emotional content. ,
distribution and project they found that commit
approval. comments written on
Mondays tend to a more
negative emotion.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
9 Sentiment Journal of the Twitter is a popular micro They used the sentiment This paper analyzed over 3,000,000
analysis of American blogging service that is used to analysis technique as a tweets mentioning or replying to
twitter Society for read and write millions of short valid popularity the 13 most influential users to
audiences: Information messages on any topic within a indicator or First, they determine audience sentiment.
Measuring the Science and 140‐character limit. Popular or distinguished between Twitter messages reflect the
positive or Technology influential users tweet their the positive and landscape of sentiment toward its
negative banner,2012 status and are retweeted, negative audiences of most popular users.
influence of mentioned, or replied to by their popular users. Second,
popular audience. Sentiment analysis of they found that the
twitterers the tweets by popular users and sentiments expressed in
their audience reveals whether the tweets by popular
the audience is favourable to users influenced the
popular users. sentiment of their
audience. Third, from
the above two findings
they developed a
positive‐negative
measure for this
influence. Finally, using
a Granger causality
analysis, they found that
the time‐series‐based
positive‐negative
sentiment change of the
audience was related to
the real‐world sentiment
landscape of popular
users.measure.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
10 Sentiment Association for Tweets are produced in many This paper offers three Developing systems able toperform
Computational languages, but most work on sentiment analysis for tweets in a new
analysis on main contributions in
Linguistics, sentiment analysis is done for language requires at least a corpus of
Italian tweets this respect. First, by
Atlanta, English (even independently of tweets and a polaritylexicon
presenting the first of
Georgia, 2013 Twitter). This is also due to the corpus of tweets for
availability of tools and Italian, built in such a
resources. way that makes it
possible to use the exact
same strategy to build
similar resources for
other languages without
any manual intervention.
Second, by deriving a
polarity lexicon for
Italian, organised by
senses, also using a fully
automatic strategy which
can replicated to obtain
such a resource for other
languages. Third, by
using the lexicon to
automatically assign
polarity to two subsets of
the tweets in the Corpus,
and evaluate results
against manually
annotated.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
11 Sentiment IEEE Xplore, They are a wide platform full They developed a Social media platforms suchas blogs,
2014 of Despite great avails, and social networking sites, content
Analysis in framework that makes it
with the increasingly enormous communities and virtual worlds are
Arabic tweets possible to analyze
number of Arabic users on the tremendously becoming one of the
Twitter comments or
most powerful sources for news,
internet, this paper deals with “Tweets” as having
markets, industries, and much more.
Arabic Sentiment Analysis. positive, negative or
Thoughts, emotions, reviews and
neutral sentiments. This
feedback, which can be used inmany
can be applied in a wide
aspects.
range of applications
ranging from politics to
marketing. This
framework has many
novel aspects such as
handling Arabic dialects,
Arabizi and emoticons.
Also, crowd sourcing
was utilized to collect a
large dataset of tweets
12 Visual IEEE This paper uses Visual sentiment To explore high-volume Twittercurrently receives about 190
Conference on analysis on twitter data streams million tweets a day, in which people
sentiment twitter data, they
VisualAnalytics for high volume twitter data share their comments regarding a
analysis on introduce three novel
Science and which are otherwise not so easy. wide range of topics. A large number
twitter data time-based visual
of tweets include opinions about
streams Technology sentiment analysis
products and services. However, with
(VAST), techniques: (1) topic-
Twitter being a relatively new
2011 based sentiment analysis
phenomenon, these tweets are
that extracts, maps, and
underutilized as a source for
measures customer
evaluating customersentiment.
opinions; (2) stream
analysis that identifies
interesting tweets based
on their density,
negativity, and influence
characteristics;
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
Production of consistent results and (3) pixel cell-based
with the large amount of data sentiment calendars and
to be processed is a difficulty high density geo maps
hence an automated opinion that visualize large
mining and summarization volumes of data in a
systems are needed to single view. They
overcome subjective biases applied these techniques
and mental limitations with an to a variety of twitter
objective sentiment analysis data, (e.g., movies,
system. amusement parks, and
hotels) to show their
distribution and patterns,
and to identify
influential opinions.

13 Survey of University of In the modern era of social media Production of consistent Twittercurrently receives about 190
Illinois at (i.e., reviews, forum discussions, million tweets a day, in which people
mining and results with the large
Chicago(2012) blogs and social networks), share their comments regarding a
sentiment amount of data to be
people or company are able to wide range of topics. A large number
analysis processed is a difficulty
get public opinions throughthese of tweets include opinions about
hence an automated
platforms for their decision products and services. However, with
opinion mining and
making.. Twitter being a relatively new
summarization systems
phenomenon, these tweets are
are needed to overcome
underutilized as a source for
subjective biases and
evaluating customersentiment.
mental limitations with
an objective sentiment
analysis system.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no. paper Year
14 A Semantic Nanyang In this paper the difference Common sense reasoning is a Senticnet focused on gathering
Resource for Technological between the AI system and good solution for sentiment information which humans
Sentiment University (11- human intelligence to harness analysis but the scalability of take for granted but were
Analysis 17 2016) common sense knowledge common sense knowledge heavily limited by their
Based on obtained from a lifetime bases is a major factor that inability to generalize.
Conceptual learning and experience in jeopardizises the efficiency of
Senticnet 4 solves this
Primitives informed decision making is concept extraction and
mentioned. Polarity detection. A first problem by leveraging on
As humans can easily adapt to possible step in solving this conceptual primitives
novel situations where AI fails problem is to generalize automatically generated by
due to lack of situation-specific pieces of common sense hierarchical clustering and
rules and capabilities. knowledge in terms of dimensionality reduction.
conceptual primitives that
could catch most semantic
inflections of natural language
concepts.
15 Who cares The University Sarcasm occurs frequently in detection of sarcasm in Working from left to right, we
about of Sheffield,( 6 user-generated content such as tweets, while useful, is not use a Viterbi-like algorithm to
sarcastic March 2014) blogs, forums and micro posts, sufficient for accurate look for the best possible
tweets? especially in English, and is sentiment analysis of such match that combines a set of
Investigating inherently difficult to analyse, tweets. Adding rules to deal known words from the
the impact of not only for a machine but even with the scope of sarcastic lookups, and completes the
sarcasm on for a human. One needs to have hashtags does, however, tokenisation to the end of the
sentiment a good understanding of the improve performance hashtag. If a combination of
analysis context of the situation, the considerably, though further matches can be found without a
culture in question, and perhaps improvements could still be break, the individual
the very specific topic or people made. components are converted to
tokens and the original single
involved in the sarcastic
statement. Furthermore, tweets
labelled with the hashtag.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
16 Sentiment Morgan and The growing importance of using microblogs and
CLaypool
Analysis and sentiment analysis coincides twitter they have tried to
publisher. 2012
Opinion with the growth of social media collect the different
Mining such as reviews, forum opinion of people about
discussions, blogs, micro-blogs, a product and then based
Twitter, and social networks. on the data they have
For the first time in human tried analyze different
history, we now have a huge sentiments and presented
volume of opinionated data in form of pie charts.
recorded in digital form for
analysis.
Sentiment analysis systems are
being applied in almost every
business and social domain
because opinions are central to
almost all human activities and
are key influencers of our
behaviors. Our beliefs and
perceptions of reality, and the
choices we make, are largely
conditioned on how others see
and evaluate the world. For this
reason, when we need to make a
decision we often seek out the
opinions of others. This is true
not only for individuals but also
for organizations. Using natural
language processing, social
media analysis, text mining, and
data mining.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
17 Recognizing Advanced This paper presents a new : based on the following
Contextual Research and approach to phrase-level features they tried to
Polarity in Development sentiment analysis that first determine the polarity of
Phrase-Level Activity determines whether an the review given by the
Sentiment (ARDA) expression is neutral or polar user
Analysis and then disambiguates the •Word token
polarity of the polar •Word part of speech
expressions. With this •Word context
approach, the system is able to •Prior polarity positive
automatically identify the negative or neutral
contextual polarity for a large •Reliability strong
subset of sentiment expressions, subject or weak subject
achieving results that are
significantly better than
baseline.
18 Twitter as a the NSF under In our paper, we focus on using •We have presenteda • We used the collected corpusto
Twitter, the most popular method for an automatic train a sentiment classifier. Our
Corpus for grant IIS-
microblogging platform, for the collection ofa corpus that
Sentiment 0208798 and classifier is able to determine
task of sentiment analysis. We can be used to train a
Analysis and by the positive, negative and neutral
sentiment classifier. We
Opinion Advanced show how to automatically sentiments of documents. The
used TreeTagger for POS-
Mining Research and collect a corpus for sentiment classifier is based on the
tagging and observed the
Development analysis and opinion mining multinomial Na¨ıve Bayes
difference in distributions
Activity purposes. We perform linguistic classifier that uses N-gram and
among positive, negative
(ARDA) analysis of the collected corpus POS-tags as features.
and neutral sets. Fromthe
and explain discovered observations we conclude
phenomena. Using the corpus, that authors use syntactic
we build a sentiment classifier, structures to describe
that is able to determine emotions orstate facts.
positive, negative and neutral Some POS-tags maybe
sentiments for a document. strong indicators of
emotional text.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
19. ASentimental Association for Sentiment analysis seeks to they have shown that
Computational
Education: identify the viewpoint(s) employing the
Linguistics
Sentiment underlying a text span; an minimum-cut framework
Stroudsburg,PA,
Analysis example application is results in the
USA©2004
Using classifying a movie review as development of efficient
Subjectivity “thumbs up” or “thumbs algorithms for sentiment
Summarizatio down”. To determine this analysis. Utilizing
n Based on sentiment polarity, they propose contextual information
Minimum a novel machine-learning via this framework can
Cuts method that applies text- lead to statistically
categorization techniques to just significant improvement
the subjective portions of the in polarity-classification
document. Extracting these accuracy
portions can be implemented
using efficient techniques for
finding minimum cuts in
graphs; this greatly facilitates
incorporation of cross-sentence
contextual constraints
20. Sentiment Handbook of 1. Feature-based sentiment •They then described •The topic of opinion search or
analysis feature-based sentiment
Analysis and Natural retrieval was introduced as well, as a
2. Sentiment analysis of analysis which exploits
Subjectivity , Language parallel to the general Web search.
the full power of the
2010 Processing, comparative sentences Last but not least, we discussed
abstract model. Afterthat
Second 3. Opinion search and retrieval opinion spam, which is increasingly
we discussed theproblem
Edition, 4. Opinion spam and utility of becoming an important issue as
of analyzing comparative
(editors: N. opinions more and more people are relying on
and superlative sentence.
Indurkhya and opinions on the Web for decision
Such sentences represent
F. J. Damerau), making. This gives more and more
a different type of
2010 incentive for spam. There is still no
evaluation from direct
opinions whichhave been
effective technique to combat
the focus of the current
opinion spam
research.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
Twitter Stanford The purpose of their project was Their hypothesis was that this type of sentiment analysis is useful
21 Sentiment 2015 to build an algorithm that can they for consumers who are trying to
Analysis accurately classify Twitter could obtain high accuracy research
messages as positive or negative, on classifying sentiment in a product or service, or marketers
with respect to a query term. Twitter messages using researching public opinion of their
machine company.
learning techniques.

Analysis in School of EECS, In this paper, instead of pre- Development of a baseline

22 Twitter: A Peking senting the sentiment polarity of approach on sentiment
Graph-based University, each tweet relevant to the topic, analysis results of the
Hashtag Beijing they focus their study on hashtag- tweets containing the
Sentiment level sentiment classification. hashtag through
Classification simple voting strategy. In
Approach order to improve the
hashtag level
sentiment classification,
use of a graph model to
boost
the results from the voting
baseline, which effectively
incorporates
the tweets sentiment
information and hashtags
co-occurrence relationship.
The preliminary results
demonstrate that our graph
model
is able
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
A survey on 2018 This paper presents a rigorous Apart from SVM, NN and This type of sentiment analysis is
23 opinion mining survey on sentiment analysis, lexicon based approaches; useful for consumers who are trying to
and sentiment which portrays views presented by we found that some of the research a product or service, or
analysis: over one hundred articles intelligent techniques have marketers researching public opinion
Tasks, published in the last decade not been exploited of their company.
approaches regarding necessary tasks, exhaustively like random
and approaches, and applications of forest, evolutionary
applications sentiment analysis computation, association
rule mining, fuzzy rule
based systems, rule miner,
conditional random field
theory (CRF), formal
concept analysis, radial
basis function neural
network (RBFNN), and
online learning algorithms.
A study of University of Most sentiment analysis for accurate
24 Information Wolverhampton approaches use as baseline a classification it is
Retrieval support vector machines (SVM) important to use term
weighting classifier with binary unigram weighting
schemes for weights. functions that scale
sentiment In this paper, they explored sublinearly in relation to
analysis whether more sophisticated the number of times a term
feature weighting schemes occurs in a document
from Information Retrieval can and that document
enhance frequency smoothing is a
classification accuracy. significant
factor.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
A System for 2015 This paper describes a system for Use of the Twitter
25 Real-time University of real-time analysis of public “firehose”and expert-
Twitter Southern sentiment toward presidential curated rules and keywords
Sentiment California candidates in the 2012 U.S. to get a full and accurate
Analysis of election as expressed on Twitter, a picture of the online
2012 U.S. microblogging service. Twitter political landscape. Use of
Presidential has become a central site where real-time data processing
Election Cycle people express their opinions and infrastructure and
views on political parties and statistical sentiment model
candidates. Emerging events or evaluates public sentiment
news are often followed almost changes in response to
instantly by a burst in Twitter emerging political events
volume, providing a unique and news as they unfold.
opportunity to gauge the relation
between expressed public
sentiment and electoral events. In
addition, sentiment analysis can
help explore how these events
affect public opinion.
Sentiment National This paper introduces an approach The method introduced in
26 analysis using Institute of to sentiment analysis which uses this paper allows several
support vector Informatics support vector machines (SVMs) methods of assigning
machines with to bring together diverse sources semantic values to phrases
diverse of potentially pertinent and words within a text to
information information, including several be exploited in a more
sources favorability measures for phrases useful way than was
and adjectives and, where previously possible, by
available, knowledge of the topic incorporating them as
of the text. features for SVM
modeling, and for explicit
topic information to be
utilized, when available, by
features incorporating such
values.
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
27 Twitter Alec Go We introduce a novel approach We show that using The idea behind Maximum Entropy
Sentiment Stanford for automatically classifying emoticons as noisy models is that one should prefer the
Classification University the sentiment of Twitter labels for training data is most uniform models that satify a
using Distant Stanford, 2013 messages. These messages are an effective way to given constraint . MaxEnt models
Supervision classified as either positive or perform distant are feature-based models.
negative with respect to a supervised learning.
query term. This is useful for Machine learning
consumers who want to algorithms (Naive
research the sentiment of Bayes, maximum
products before purchase, or entropy classification,
companies that want to and support vector
machines) can achieve
monitor the public sentiment of
high accuracy for
their brands.
classifying sentiment
when using this method.
28 Building ACM New York, Recognizing polarity requires a Sentiment analysis is a The basic idea is exploiting certain
Lexicon for NY, USA,2014 list of polar words and phrases. recent attempt to deal language and layout structures as
Sentiment For the purpose of building with evaluative aspects clues to extract polar sentences. The
Analysis from such lexicon automatically, a of text. In sentiment clues were carefully chosen so that
Massive lot of studies have investigated analysis, one it achieves high precision.
Collection of (semi-) unsupervised method fundamental problem is
HTML of learning polarity of words to recognize whether
Documents and phrases. In this paper, we given text expresses
explore to use structural clues positive or negative
that can extract polar sentences evaluation. Such
from Japanese HTML property of text is called
polarity
documents, and build lexicon
from the extracted polar
sentences
S Title of the Journal & Abstract Significant Critical Remarks
no. paper Year Outcomes
29 Subjectivity Journal of the Although Subjectivity and The only work onArabic In this paper, we build a sentence-
and Sentiment American Sentiment Analysis (SSA) has SSA we are aware of is level SSA system for MSA
Analysis of Society for been witnessing a flurry of that of Abbasi et al. contrasting language independent
Modern Information novel research, there are few (2008). They use an only features vs. combining
Standard Science and attempts to build SSA systems entropy weighted language independent and
Arabic Technology for Morphologically-Rich genetic algorithm for language-specific feature sets,
banner,2012 Languages (MRL). In the both English and Arabic namely morphological features
current study, we report efforts Web forums at the specific to Arabic. We also
to partially fill this gap. We document level. investigate the level of stemming
present a newly developed required for the task. We show that
manually annotated corpus of the Stem lemmatization setting
Modern Standard Arabic outperforms both Surface and
(MSA) together with a new Lemma settings for the SSA task.
polarity lexicon..The corpus is
a collection of newswire
documents annotated on the
sentence level
30 Stock Association for In this paper, we apply We have investigated the Granger causality is based on linear
Computational sentiment analysis and machine causative relation regression, but the correlation
Prediction
Linguistics, learning principles to find the between public mood as between stocks and moods is
Using Twitter
Atlanta,Georgia, correlation between ”public measured from a large certainly non linear. Therefore, after
Sentiment
2013
Analysis sentiment” and ”market scale collection of tweets finding a causality relation between
sentiment”. We use twitter data from twitter.com and the the past 3 days moods and current
to predict public mood and use DJIA values. Our results day stock prices
the predicted mood and previous show that firstly public
days’ DJIA values to predict the mood can indeed be
stock market movements. captured from the large-
scale Twitter feeds by
means of simple natural
language processing
techniques, as indicated
by the responses.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no. paper Year
31 Cross- IEEE Xplore, We explore the adaptation of The surge in attention paid to For calculation of semantic
Linguistic 2012 English resources and automated analysis of text sentiment orientation using lexicons,
Sentiment techniques for text sentiment has largely been focused on English. In translation of any kind
Analysis: analysis to a new language, this paper, we have discussed how to seems to come with a price,
From English Spanish. Our main focus is the adapt an existing English semantic even between closely
to Spanish modification of an existing orientation system to Spanish while at related languages such as
English semantic orientation the same time comparing several English and Spanish.
calculator and the building of alternative approaches
dictionaries; however we also
compare alternate approaches,
including machine translation
and Support Vector Machine
classification.

32 sentiment IEEE Xplore: This paper uses Visual To explore high-volume twitter data, Granger causality is based
analysis on 30 January sentiment analysis on twitter they introduce three novel time-based on linear regression, but
twitter using 2014 data streams for high volume visual sentiment analysis techniques: the correlation between
machine twitter data which are (1) topic-based sentiment analysis that stocks and moods is
learning otherwise not so easy. extracts, maps, and measures customer certainly non linear.
techniques. opinions; (2) stream analysis that Therefore, after finding a
identifies interesting tweets based on causality relation between
their density, negativity, and influence the past 3 days moods and
characteristics; and (3) pixel cell-based current day stock prices
sentiment calendars and high density
geo maps that visualize large volumes
of data in a single view. They applied
these techniques to a variety of twitter
data, (e.g., movies, amusement parks,
and hotels) to show their distribution
and patterns, and to identify influential
opinions.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no. paper Year
33 Sentiment IEEE (May- This paper focuses on In this page, the main focus is on
Analysis in an June 2013) analysing speaker’s sentiment automatically analysing speaker’s
Audio-Visual through videos consisting of sentiment in online videos by adding
Context movie reviews. This approach audio feature that is used in speech-
adds audio features used based emotion recognition as well as
speech-based emotion video features along with the textual
recognition as well as video information. By this, the sentiment of
features encoding valuable the speaker is conveyed for effectively
information delivered by the and efficiently.
speaker.
34 Fine-Grained University of Sentiment analysis is the Markov logic integrate polarity scores Sentiment analysis is the
Sentiment Mannheim, problem of determining the from different sentiment lexicons with problem of determining the
Analysis with Germany(201 polarity of a text with respect information about relations between polarity of a text, it is also
Structural 2) to a particular topic. neighbouring segments, and evaluate difficult to extract the
Features Sentiment analysis systems the approach on product reviews. The negative or positive
working on the (sub-)sentence experiments show that the use of utterances on a more fined
level, however, are difficult to structural features improves the grained level. In this paper,
develop since shorter textual accuracy of polarity predictions a fully automatic
segments rarely carry enough achieving accuracy scores of up to 69% framework for fine-grained
information to determine their sentiment analysis on the
polarity out of context. In this sub-sentence level
paper, therefore, we present a combining multiple
fully automatic framework for sentiment lexicons and
fine-grained sentiment neighbourhood as well as
analysis on the sub-sentence discourse relations to
level combining multiple overcome the problem.
sentiment lexicons and
neighbourhood as well as
discourse relations to
overcome this problem.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no paper Year
.
35 Sentic patterns: The Scientific In this work, we introduce a By allowing sentiments to flow from
Dependency- World Journal novel paradigm to concept- concept to concept based on the
based rules for Volume 2011, level sentiment analysis that dependency relation of the input
concept-level Article ID 634396 merges linguistics, common- sentence, in particular, we achieve a
sentiment sense computing, and machine better understanding of the
analysis learning for improving the contextual role of each concept
accuracy of tasks such as within the sentence and, hence,
polarity detection. obtain a polarity detection engine
that outperforms state-of-the-art
statistical methods.

36 The impact of IEEE This study aims to investigate We apply the advanced sentiment
social and Department of the effect of social media and analysis technique that goes beyond
conventional Computer Science conventional media, their the number of mentions (counts) to
media on firm Tsuda College relative importance, and their analyze the overall sentiment of
equity value: A ,2006 interrelatedness on short term each media resource toward a
sentiment firm stock market specific company on the daily basis.
analysis performances. We use a novel We use stock return and risk as the
approach and large-scale dataset that indicators of companies' short-term
features daily media content performances.
across various conventional
media and social media
outlets for 824 public traded
firms across 6 industries.
Social media outlets include
blogs, forums, and Twitter.
Conventional media includes
major newspapers, television
broadcasting companies, and
business magazines.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no paper Year
.
37 A Preliminary Department of With the rise of weblogs and We describe preliminary statistical
Investigation Linguistics and the increasing tendency of tests on a new dataset of political
into Sentiment Oriental online publications to turn to discussion group postings which
Analysis of Languages San message-board style reader indicate that posts made in direct
Informal Diego State feedback venues, informal response to other posts in a thread
Political University political discourse is have a strong tendency to represent
Discourse becoming an important an opposing political viewpoint to
feature of the intellectual the original post
landscape of the Internet,
creating a challenging and
worthwhile area for
experimentation in techniques
for sentiment analysis.

38 Adapting Naive Key Laboratory In the community of This is so-called domain-transfer To leverage knowledge
Bayes to of Network sentiment analysis, supervised problem. In this work, we attempt to from the old-domain data,
Domain Institute of learning techniques have been attack this problem by making the we proposed an effective
Adaptation for Computing shown to perform very well. maximum use of both the old- measure, i.e., Frequently
Sentiment Technology When transferred to another domain data and the unlabeled new- Co-occurring Entropy
Analysis ,China domain, however, a domain data. (FCE), to pick out
March2009 supervised sentiment classifier generalizable features that
often performs extremely bad. occur frequently in both
domains and have similar
occurring probability.
S Title of the Journal & Abstract Significant Outcomes Critical
no paper Year Remarks
.
39 How Do Users Technische App stores allow users to submit for many apps, the amount of
Like This Universität feedback for downloaded apps in reviews is too large to be processed
Feature? A Fine München, form of star ratings and text manually and their quality varies
Grained Garching, reviews. Recent studies analyzed largely. The star ratings are given to
Sentiment Germany this feedback and found that it the whole app and developers do not
Analysis of App includes information useful for app have a mean to analyze the feedback
Reviews developers, such as user for the single features.
requirements, ideas for
improvements, user sentiments
about specific features, and
descriptions of experiences with
these features.
40 Explicit factor Tsinghua In this work, we propose the Offline experimental results on
models for University Explicit Factor Model (EFM) to several real-world datasets
explainable 11 July 2014 generate explainable demonstrate the advantages of our
recommendatio recommendations, meanwhile keep framework over competitive baseline
n based on a high prediction accuracy. We first algorithms on both rating prediction
phrase-level extract explicit product features and top-K recommendation tasks.
sentiment (i.e. aspects) and user opinions by
analysis phrase-level sentiment analysis on
user reviews, then generate both
recommendations and
disrecommendations according to
the specific product features to the
user's interests and the hidden
features learned. Besides,
intuitional feature-level
explanations about why an item is
or is not recommended are
generated from the model.
S Title of the Journal & Abstract Significant Outcomes Critical
no paper Year Remarks
.
41 International Dept. of There is a growing interest in Our experiments indicate that (a)
Sentiment Computer Science mining opinions using sentiment entity sentiment scores obtained by
Stony Brook analysis methods from sources our method are statistically
Analysis for
University such as news, blogs and product significantly correlated across nine
News and Blogs
reviews. Most of these methods languages of news sources and five
have been developed for English languages of a parallel corpus; (b)
and are difficult to generalize to the quality of our sentiment analysis
other languages. We explore an method is largely translator
approach utilizing state-of-the-art independent; (c) after applying
machine translation technology and certain normalization techniques,
perform sentiment analysis on the our entity sentiment scores can be
English translation of a foreign used to perform meaningful cross-
language text. cultural comparisons.

42 Automatic Conference on The growing stream of content A sentiment isn’t always stated in a
Sentiment Electronic placed on the Web provides a huge clear way in the text; it is often
Analysis in On- Publishing – collection of textual resources. represented in subtle, complex
line Text Vienna, Austria – People share their experiences on- ways. Besides direct expression of
June 2007 line, ventilate their opinions (and the user's feelings towards a certain
frustrations), or simply talk just topic, he or she can use a diverse
about anything. The large amount range of other techniques to express
of available data creates his or her emotions.
opportunities for automatic mining
and analysis. The information we
are interested in this paper, is how
people feel about certain topics. We
consider it as a classification task:
their feelings can be positive,
negative or neutral. .
S Title of the Journal & Abstract Significant Outcomes Critical
no paper Year Remarks
.
43 Subjectivity and Portland, Oregon Although Subjectivity and In the current study, we report
sentiment June 24, 2011 Sentiment Analysis (SSA) has been efforts to partially fill this gap. We
analysis of witnessing a flurry of novel present a newly developed manually
modern research, there are few attempts to annotated corpus of Modern
standard build SSA systems for Standard Arabic (MSA) together
Morphologically-Rich Languages with a new polarity lexicon.
(MRL).

44 Twitter Stanford We introduce a novel approach for There is no previous research on

Sentiment University automatically classifying the classifying sentiment of messages
Classification Stanford sentiment of Twitter messages. on microblogging services like
using Distant These messages are classified as Twitter. We present the results of
Supervision either positive or negative with machine learning algorithms for
respect to a query term. This is classifying the sentiment of Twitter
useful for consumers who want to messages using distant supervision.
research the sentiment of products Our training data consists of Twitter
before purchase, or companies that messages with emoticons, which are
want to monitor the public used as noisy labels.
sentiment of their brands.
S Title of the Journal & Abstract Significant Outcomes Critical
no paper Year Remarks
.
45 Sentiment Technische With the development of Web 2.0, The local conjunctive words, such most of them make
Analysis with Universität sentiment analysis has now become as “and” or “but”, are often the assumption that,
München, a popular research problem to indicative of sentiment transitions. given the parameters,
Global Topics
Garching, tackle. Recently, topic models have In this paper, we propose a major the sentiments of the
and Local Germany been introduced for the departure from the previous words in the
Dependency simultaneous analysis for topics approaches by making two linked document are all
and the sentiment in a document. contributions. First, we assume that independent. In our
These studies, which jointly model the sentiments are related to the observation, in
topic and sentiment, take the topic in the document, and put contrast, sentiments
advantage of the relationship forward a joint sentiment and topic are expressed in a
between topics and sentiment, and model, i.e. Sentiment-LDA coherent way.
are shown to be superior to
traditional sentiment analysis tools.

46 Explicit factor Tsinghua In this work, we propose the Besides, intuitional feature-level
models for University Explicit Factor Model (EFM) to explanations about why an item is
explainable 11 July 2014 generate explainable or is not recommended are
recommendatio recommendations, meanwhile keep generated from the model. Offline
n based on a high prediction accuracy. We first experimental results on several real-
phrase-level extract explicit product features world datasets demonstrate the
sentiment (i.e. aspects) and user opinions by advantages of our framework over
analysis
phrase-level sentiment analysis on competitive baseline algorithms on
user reviews, then generate both both rating prediction and top-K
recommendations and recommendation tasks.
disrecommendations according to
the specific product features to the
user's interests and the hidden
features learned.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no paper Year
.

47 GERMANPOL LREC, 2010 we propose German Polarity we proposed a new publicly However, the current study
Clues, a new publicly available available lexical resource for can only be seen as a starting
ARITYCLUES:
lexical resource for sentiment sentiment analysis for the point in the construction of
A Lexical
analysis for the German German language – German resources for a German
Resource for language. While sentiment Polarity Clues. The new based sentiment analysis.
German analysis and polarity resource was built combining a Future work includes the
Sentiment classification has been semi-automatic translation extension and revalidation of
Analysis extensively studied at different method and a manually the existing dataset with
document levels (e.g. sentences assessment and extension of additional polarity features
and phrases), only a few individual polarity-based term as aggregated from other
approaches explored the effect features. We empirically showed (webbased) resources and
of a polarity-based feature that the German Polarity Clues dictionaries. We also plan to
selection and subjectivity dictionary can be, with an F1- conduct an human-
resources for the German Measure of 87.6, a valuable judgement-based assessment
language. This paper evaluates resource for a polarity-based of the other two resources, in
four different English and three feature selection order to improve the existing
different German sentiment German Polarity Clues
resources in a comparative dictionary
manner by combining a
polarity-based feature selection
with SVM-based machine
learning classifier.
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no paper Year
.

48 Improving Elsevier,2017 propose a divide-and-conquer The approach employs However, the current study
sentiment approach which first classifies BiLSTM-CRF to extract target can only be seen as a starting
analysis via sentences into different types, expression in opinionated point in the construction of
sentence type then performs sentiment sentences, and classifies these resources for a German
classification analysis separately on sentences sentences into three types based sentiment analysis.
using BiLSTM- from each type. Classifying according to the number of Future work includes the
CRF and CNN
sentence types by the number of targets extracted from extension and revalidation of
opinion targets a sentence them.Empirical results show that the existing dataset with
contain. our approach achieves state-of- additional polarity features
A data-driven approach the-art performance on three of as aggregated from other
automatically extract features the four datasets. We have found (webbased) resources and
from input sentences. that separating sentences dictionaries. We also plan to
containing different opinion conduct an human-
targets boosts the performance judgement-based assessment
of sentence-level sentiment of the other two resources, in
analysis. order to improve the existing
German Polarity Clues
dictionary
S Title of the Journal & Abstract Significant Outcomes Critical Remarks
no. paper Year
49 SENTIWORD LREC,2010 We here discuss Two different random-walk
NET 3.0: An SENTIWORDNET 3.0, processes are executed for the
Enhanced especially focussing on the positivity and negativity
Lexical improvements concerning aspect dimensions, respectively, of
Resource (b) that it embodies with respect SENTIWORDNET, producing two
for Sentiment to version 1.0. We also report the different rankings of the
Analysis and results of evaluating WORDNET synsets. However, the
Opinion SENTIWORDNET 3.0 against a actual numerical values
Mining fragment of WORDNET 3.0 returned by the random-walk
manually annotated for process are unfit to be used as the
final Pos and Neg scores, since they
Positivity, negativity, and
are all too small as a result, even
neutrality; these results indicate
the top-ranked positive synsets
accuracy improvements of about
would turn out to be
20% with respect to overwhelmingly neutral and only
SENTIWORDNET 1.0. feebly positive.
50 LargeScale LCWSM,2017 They present a system that There are many interesting
Sentiment assigns scores indicating directions that can be explored.
Analysis for positive or negative opinion to We are interested in how sentiment
News and each distinct entity in the text can vary by demographic
Blogs corpus. Their system consists of group, news source or geographic
a sentiment identication phase, location. By expanding our
which associates expressed spatial analysis of news entities to
opinions with each relevant sentiment maps, we can
entity, and a sentiment identify geographical regions of
aggregation and scoring phase, favorable or adverse opinions
which scores each entity relative for given entities. We are also
to others in the same class. studying in analyzing the de-
Finally, we evaluate the gree to which our sentiment indices
significance of our scoring predict future changes in
techniques over large corpus popularity or market behavior.
of news and blogs
Knowledge Gained
• Various approaches of how to identify the key critical
remarks over a sentiment was validated in the papers.
• The scope of extending the outlooks from specificreviews
to twitter tweets can also be used as a way to collect the
data.
• In one of the papers, there key way to analyze sentiments
were to use automated opinion mining and use it to detect
sarcasm.
• First possible step in solving any problem related to the
issue is to generalize pieces of common sensesknowledge
in terms of conceptual primitives that could catch most
semantic inflections of natural languageconcepts.
Knowledge Gained
• Determining the polarity of a text, it is also
difficult to extract the negative or positive
utterances on a more fined grained level, soit’s
important to identify such words while doing
word minning.
• Various probablity theorems were taken in
considertation to predict the nature of
sentiments.
• Use of weighted scheme from information
retrieval to enhance accuracy of sentimental
analysis
Problems Identified
• In the paper it was mentioned that spam reviews
are not accountable as a relevant data set for the
sentimental analysis.
• Language barrier, reviews in different languages
can’t be analyzed.
• It is hard to identify sarcasm without the
particular set of trained natural language
processing identifier.
• The polarity of the review is not easy to be
identified due to complicatedlanguage.
Motivation for the ResearchWork
• After facing the same problem day in and day out over
where to buy the product from, it is needed to make a
standardize effort on identifying what can be the
determining parameter to address the problem. This
project intends to deal with the fore-mentioned problem.
• The customer reviews on websites of amazon, flipkart, snap
deal and many more, post their thoughts on the product
delivery.
• It is hard to go through all the text and reviews to identify
the results one is looking for.
• Sentiment analysis gives a list of word frequencies that is
repeated in a set of text, which gives a basic outlook on
which a ground level conclusion can be made.
Objective and Scope
• The main objective of this project is to create
a platform that is unbiased of the
predicaments of the buyers who paywebsites
to show good ratings of thereproducts.
• This type of analysis opens the scope for,call
recording and analysing the recorded
conversation and predicting about the
customers interest in a particular product.
Work Plans
S.No Task Time-Taken
1) Data Collection 1-week
2) Processing of data 1-week
3) Sentiment Analysis on data 10 days
4) Conclusions and results 1-week
5) Drafting report 1-week
6) Project Submission 2-weeks
TARP REVIEW 2
PRESENTED BY GROUP 2
SENTIMENTAL ANALYSIS
ON IPHONE
GROUP MEMBERS

 16BME2024- AYUSH AGARWAL (7708883510)

 16BME0341- ANIRUDH SHANKAR (9495368881)
 16BME0598- GULSHAN KUMAR (8667254252)
 16BME0756- HARSHIT MEHTA (8053600036)
 16BPI0047- AASHRAY NAGRATH (8288987052)
 16BPI0030- KUSHAGRA TIWA (7530000496)
 16BME2031- PRAKHAR CHAWLA (7417467321)
 16BME0930- PARAS MUKHIJA (9751666841)
 16BME0767- KARTIK MATHUR (8003912839)
OBJECTIVE

 The main objective of the project is to create a platform that is

unbiased of the predicaments of the buyers who pay websites to
show good ratings of there products.
 This type of analysis opens the scope for, call recording and
analysing the recorded conversation and predicting about the
customer interest in a particular product.
Standards

 TEXT MINNING- Text mining, also referred to as text data mining,

roughly equivalent to text analytics, is the process of deriving high-
quality information from text. High-quality information is typically
derived through the devising of patterns and trends through means
such as statistical pattern learning.
 WORD CLOUD- A word cloud is a novelty visual representation of
text data, typically used to depict keyword metadata on websites,
or to visualize free form text
METHODOLOGY

DATA
COLLECTION

TEXT MINING

WORD
CLOUD
FORMATION
METHODOLOGY EXPLAINED

 First step was data collection where we scraped data from Amazon
website. It contained reviews from customers on iphone 10 and one
plus 6 product delivery.
 Next we used R code software to do text mining where frequently
appeared words were tabulated.
 The tabulated words were used to plot word cloud to get a pictorial
representation of the result.
 Analysing the word cloud, we could analyse the better product
delivery among the two flagship companies.
Software Used

R is a programming language and free software environment for

statistical computing and graphics supported by the R Foundation for
Statistical Computing. The R language is widely used among
statisticians and data miners for developing statistical software and
data analysis
Experiment Performed

 Collected data from Amazon regarding the unbiased opinion on

two products.
 Used text mining on the collected data using the R code.
 Displayed the word frequency, after doing the text mining.
 Word cloud generated.
Source of data collection
https://www.amazon.in/Apple-iPhone-
Space-Grey-Storage/product-
reviews/B072LPF91D
Data Collected
Data Collected
Data Collected
Data Collected
Data Collected
R code used
 CODE:
 library("tm")
library("SnowballC")
library("wordcloud")
library("RColorBrewer")
text <- readLines(file.choose("IPHONE 10 PRODUCT REVIEW.txt"))
docs <- Corpus(VectorSource(text))
inspect(docs)
toSpace <- content_transformer(function (x , pattern ) gsub(pattern, " ", x))
docs <- tm_map(docs, toSpace, "/")
docs <- tm_map(docs, toSpace, "@")
docs <- tm_map(docs, toSpace, "\\|")
docs <- tm_map(docs, content_transformer(tolower))
docs <- tm_map(docs, removeNumbers)
docs <- tm_map(docs, removeWords, stopwords("english"))
docs <- tm_map(docs, removeWords, c(“iphone", “phone"))
docs <- tm_map(docs, removePunctuation)
docs <- tm_map(docs, stripWhitespace)
docs <- tm_map(docs, stemDocument)
dtm <- TermDocumentMatrix(docs)
m <- as.matrix(dtm)
v <- sort(rowSums(m),decreasing=TRUE)
d <- data.frame(word = names(v),freq=v)
head(d, 10)
set.seed(1234)
wordcloud(words = d$word, freq = d$freq, min.freq = 1,
max.words=700, random.order=FALSE, rot.per=0.35,
colors=brewer.pal(8, "Dark2"))
WORKING MODEL
WORKING MODEL
RESULTS AND OUTCOMES
Oneplus

word frequency
one 33
purchase 28
quality 28
verified 25
device 23
good 23
iphone 23
charge 20
battery 19
camera 19

IPHONE

WORD FREQUENCY
PURCHASE 35
COLOUR 25
VERIFIED 25
AMAZON 17
APPLE 15
SILVERS 14
SPACE 12
GREY 11
DAY ONE 10
RESULTS
Word cloud of
I phone:
RESULTS
Word cloud of
One plus:
RESEARCH OUTCOME
 Our work is titled as SENTIMENTAL ANALYSIS OF IPHONE AND ONE PLUS PRODUCT
DELIVERY.
 From the results obtained, it was found that one plus is better than iphone as the
number of positive reviews received for one plus are more in number. The word
good was more frequent in the one plus review.
 We were also able to identify that most of the problems incurred by Iphone were
due to miss management in the supply chain network.
 Customers had negative response towards product damage and defective
accessories.
 According to customers in comparison with iPhone one plus had better reviews
because of the impeccable battery life and no complaints over damaged
product delivery
TARP REVIEW 3
PRESENTED BY GROUP 2
SENTIMENTAL ANALYSIS
ON IPHONE
GROUP MEMBERS

 16BME2024- AYUSH AGARWAL (7708883510)

 The main objective of the project is to create a platform that is

 TEXT MINNING- Text mining, also referred to as text data mining,

DATA
COLLECTION

TEXT MINING

WORD
CLOUD
FORMATION
METHODOLOGY EXPLAINED

R is a programming language and free software environment for

 Collected data from Amazon regarding the unbiased opinion on

word frequency
one 33
purchase 28
quality 28
verified 25
device 23
good 23
iphone 23
charge 20
battery 19
camera 19

IPHONE

WORD FREQUENCY
PURCHASE 35
COLOUR 25
VERIFIED 25
AMAZON 17
APPLE 15
SILVERS 14
SPACE 12
GREY 11
DAY ONE 10
RESULTS
Word cloud of
I phone:
RESULTS
Word cloud of
One plus:
RESULTS
RESULTS
RESULTS

 The first result of the study gives us a word frequency of products.

This can be visually seen in the above figure. So here we see the
words that have been repeated the maximum number of times. This
has been done by the system. The system has recognized the
number of times a word has been used and portrayed it in the
manner of a list. We see that for the One plus phone word quality
has been used several times and also verified. This means that the
customers seem pleased by the product as they seem to like the
quality of the phone and are satisfied. The attributes like good and
verified show that the customers are satisfied and that the product is
good. This also tells the manufacturer a feedback that his products
quality is good and he needs to continue maintaining these
standards for gaining customer loyalty.
RESEARCH OUTCOME
 Our work is titled as SENTIMENTAL ANALYSIS OF IPHONE AND ONE PLUS PRODUCT
DELIVERY.
 From the results obtained, it was found that one plus is better than iphone as the
number of positive reviews received for one plus are more in number. The word
good was more frequent in the one plus review.
 We were also able to identify that most of the problems incurred by Iphone were
due to miss management in the supply chain network.
 Customers had negative response towards product damage and defective
accessories.
 According to customers in comparison with iPhone one plus had better reviews
because of the impeccable battery life and no complaints over damaged
product delivery
Conclusion

 We learnt various approaches of how to identify the key critical remarks

over a sentiment was validated in the papers. The scope of extending
the outlooks from specific reviews to twitter tweets can also be used as
a way to collect the data. In one of the papers, there key way to
analyze sentiments were to use automated opinion mining and use it to
detect sarcasm. First possible step in solving any problem related to the
issue is to generalize pieces of common senses knowledge in terms of
conceptual primitives that could catch most semantic inflections of
natural language concepts. Determining the polarity of a text, it is also
difficult to extract the negative or positive utterances on a more fined
grained level, so it’s important to identify such words while doing word
mining. Various probability theorems were taken in consideration to
predict the nature of sentiments. Use of weighted scheme from
information retrieval to enhance accuracy of sentimental analysis.
MEE3999
TARP PROJECT
SentimentAnalysis

Project Guide: Prof .S Senthil Kumaran

GROUP MEMBER
16BME2024- AYUSH AGARWAL (7708883510)
• 16BME0341- ANIRUDH SHANKAR (9495368881)
• 16BME0598- GULSHAN KUMAR (8667254252)
• 16BME0756- HARSHIT MEHTA (8053600036) 16BPI0047- AASHRAY
NAGRATH (8288987052)

• 16BPI0030- KUSHAGRA TIWARI (7530000496)

• 16BME2031- PRAKHAR CHAWLA (7417467321)
• 16BME0930- PARAS MUKHIJA (9751666841)
• 16BME0767- KARTIK MATHUR (8003912839)

DECLARATION

1|Page
This thesis is a presentation of our original research work. Wherever contributions of
others are involved, every effort is made to indicate this clearly, with due reference to
the literature, and acknowledgement of collaborative research and discussions. The
work was done under the guidance of Professor S Senthil Kumaran, at the Vellore
Institute of Technology, Vellore.

[KUSHAGRA TIWARI- Group Reprentative]

In my capacity as supervisor of the candidate’s thesis, I certify that the above

statements are true to the best of my knowledge.

[Professor S Senthil Kumaran]

Date:

2|Page
List of Table and Images:

1. Table 1- Literature Review

2. Figure 1: A Product Review from Amazon

3. Figure 2: Is A Tweet On Twitter Which Addresses Negative Review of I-phone

4. Figure 3: Spared User Comments

5. Figure 4: R code Used

6. Figure 5: word frequency generated

7. Fig: 6- word cloud from IPhone and One plus

8. Fig: 7- word cloud from Iphone and One plus

9. Fig 8- the source of source website used is amazon

10. Fig 9 - word frequency generated

11. Fig: 10- iPhone world cloud from its reviews

12. Fig: 11- word cloud from One plus

3|Page
INDEX

CONTENT PAGE NO.

1-ABSTRACT 3

2-INTRODUCTION 4-12

3-LITERATURE REVIEW 13-31

4-EXPERIMENTAL PAGE 32-37

5-RESULTS AND DISCUSSION 38-64

6-CONCLUSION AND SCOPE FUTURE WORK 65-66

7-REFERENCES 67-71

4|Page
1.ABSTRACT

KEYWORDS: sentiment analysis, product reviews, tokens, tweets, experimental,

economy

5|Page
2.Introduction

What do people do before purchasing anything? Whether it is shoes, electronic gadgets ,

foreign tour, ordering without gathering information about similar product available in the
market. This has been made easier by providing the feature of comments on online shopping
websites, online forums[28]. It gives an adequate idea to the customer if the product or service
is worth of the money. Customer prefer to involve in doing business or creation of brand
image totally depends on the degree of feedback or review by public. According to a survey
it is known that more than 90% of customer’s buying decision are influenced by the public
reviews. People don’t prefer any product or service if they receive some bad reviews about it.

The main problem with it is that people don’t usually check about the authenticity of negative
or positive review.

People share their views and opinions on social media. These views can be collected and
systematically arranged into a dataset. By analysis of this huge data, information about an
event (a current/past scenario, whether it is/was successful or not) can be identified[3][16]. The
views shared on the platform can be categorized as positive, negative or neutral. This is a
bird’s eye view of “Polarity”, based on such classification we can arrive on a conclusion
concerning an event. Since more and more people are moving to social media platforms to
express their views, Sentiment Analysis has proven to be effective in order to view what
people think about current ongoing scenarios across countries, or if considering a larger
portion of map, worldwide. It can also find its usage in companies changing their business
models, in order to have more positive sales generated from market. For this project, we
divided our workflow-into 3sub–tasks:

• Extraction of data using twitter (Dataset development): This requires the twitter API, for
extraction of tweets from the twitter database[5]. This will be the initial step for developing
the dataset, which will be further used for analysis purpose

• Analyzing the polarity and subjectivity of the dataset: After obtaining the dataset, we’ll
analyze the dataset for obtaining the polarity and subjectivity for our dataset. These are the
matrices that are defined for tracking the sentiment involved in a tweet.

6|Page
Visualization of the obtained Data: In order to get a better insight of the obtained data one has
to visualize the data, keeping our search as center point, what are the other search points or
words that are also going popular with our search. We’ll also obtain new mathematical results.
Sentimental analysis is directly linked to a person’s feeling. It can depend on a lot of
factors that may change by time such as:-

• Attitude- Attitudes are measures of ideas, object and people who suffer throughout their
life (e.g., “capital punishment is bad”). Attitudes are important because they can guide
thought, behaviour, and feelings. Attitude change occurs anytime an attitude is
modified. Thus, change takes place when a person undergoes from being pessimistic
to optimistic, from being little positive to very positive, or from no attitude to having
some. Due to the functional value of attitudes, the processes that convert them have
been a major factor from the past to present of social psychology.

• Emotion- There are many kind of emotions which have an impact on how the person
live and communicate with the surrounding people. It sometimes controls our choices,
actions and the perceptions at any time of life. Different types of emotion are:-

Basic emotion: such emotions are happiness, fear, surprise, and anger.

Combining emotion: such as joy, trust, relief, guilt shame, satisfaction.

• Opinion- it is a type of message, belief or judgement which is not based on any kind of
proof or certainty but is expressed with confidence. It is shared among people in order
to take an decision or before performing any action . it is result of emotion or
assumption. Opinion not get changed easily until or unless any new arguments [13]. It
differs from person to person as each and every individual have different perspectives,
feelings and choices.

*Semantic orientation:

* Corpus based: it requires huge amount of data to detect the polarity of the word
and hence sentiment of the data. Problem with this is that it depends on the polarity of
words that appear in training corpus as polarity is calculated for the words that are in

7|Page
the corpus. It is a simple approach. This method searches terms that consists of
sentiments from the texts and calculates the polarity. For example: “good movie”, “nice
quality”, “amazing dialogues”.

* Sentence-level lexical based: this method selects the objective and subjective
lines from different comments, blogs and reviews. The polarity then can be calculated
using semantic score as neutral, positive or negative. This method achieve efficiency of
97.8% and 86.6% at feedback and sentence level respectively.

Tools used for the process:

• Clarabridge is a management platform having customer experience. It separates and

analyse terms from chat box, reviews or blogs. It can also collect data from emails ,
call recordings , voice notes and videos. This system change the data collected to text
form.

• IBM Watson natural language consist a set of analytic system. It helps to pull data such
as important features, keywords, categories etc. emotion of the writer can vary along
with the document. It provides analysis support in more than 10 languages.

• Discover text is based on cloud having collaboration with analytic system that is helpful
for businessmen, research person and government agencies. It is useful for obtaining ,
filtering and examining data from different origins such as emails, twitter, amazon,
flipkart.

• Inmoment offers group of articles in order to make platform having customer based
experience. Facilities allow business to gather feedback from customers in the form of
calls, messages or videos. The amount of data origins are enough and consists of social
media, questionnaires.

• Microsoft text analytics API can be used to pull out important terms, things, sentiments
which can be made in 120 languages. It returns scored scoring from 0 to 1. Especially
this software can gather sentiments in languages such as French, Spanish and English.

This process of examining can be performed on entire document.

8|Page
Sentimental Analysis uses Natural Language Processing in order to analyse online
social conversations and understand its hidden meaning. NLP has the power to
understand human expressions that is fitted with computer programs. It is associated
with Artificial Intelligence. As NLP requires communication with human in a
programming language in a specific manner, it is difficult to develop NLP applications.
Very Often it is difficult to understand human expressions as it is confusing, might be
in different languages, including abusive words, regional slangs and dialects. NLP uses
two types of techniques such as semantic analysis and syntax. Semantic uses the literal
meaning of the words and applies algorithm to better understand the sentence. Other
techniques used along with semantics are resolving confusions, identification and
conversion to natural language by using database of words. Currently NLP works on
deep learning which is an integral part of AI. A huge amount of labelled data is required
by deep learning model to recognize and improve on correlations. Commonly used NLP
tools are Gensim, NTLK, Natural Language Tool Kit And Intel NLP Architect.

Syntax refers to arranging of words within a sentence to provide a sense of grammar to

it. Use of syntax by NLP is done to assess the meaning based on grammar. Some of
Syntax techniques are Division of big parts of sentences to smaller ones, placement of
boundaries in between bigger texts, division of words into groups. NLP offers many
benefits such as:

• Decreased errors and better speed while documentation Generation of summary,

which is easy to comprehend.

• Which makes it useful for personal assistants such as google assistance.

• Provides organisations for the use of chatbots during customer support
• Sentimental analysis can be performed easily using NLP

So, what are product reviews? In electronic commerce, product review is the basic and
easiest way to let the customers decide which product is better and why by providing
rates and comment on the products that they have used right on the project page. These
will be helpful for further purchases by letting other customers choose what’s better

9|Page
from the comments given by actual people who have used it. Steps to write a non
professional product review:

1. Collect knowledge about the product: To comment on the product one must
have clear ideas on what the product is. This can be done by going through
manufacturer’s websites as well as any manufacturing information about the
product. By looking into the products competing that would be beneficial in
comparing and choosing the better product.

2. Attain the product: one need to have more than just knowledge from papers
but also have a hands-on experience to truly comment on the product as it
might be different from what the person expects it to be [8]. This can be done
by either purchasing the product or renting it. It can be bought directly from
the manufacturer or from any other retailers.

10 | P a g e
3. Using the product: The real scenario or the quality of the product is known only after it
has been used. As the product are used by the reviewers, the readers can assume their
reviews to be a genuine one. It can be provided along with a photograph or a video of it
which maybe using the product like many unboxing videos that are a trend now. It can
be done to demonstrate about how the product can be used.

4. Know your Audience: It is important to know what the readers are expecting from a
review. One need to mention if the product is easy to use; the quality of the product;
advantages and disadvantages of the products and its worth; and the product’s
compatibility. For example an old person might be looking for features like larger
buttons and rigidity when it comes to buying a phone whereas for a youngster, the
performance and speed are two prime factors that they look forward to.

Some reviews are made by professionals such as bloggers and professional reviewers
who have millions of fan followers and take their words much more seriously. But often
these reviews won’t be so straight forward as they can’t seem biased to a company.
These types of reviews include:

1. Providing basic information such as name of the brand, model, dimensions of the product
and price.

2. Detailed explanation about the functionality of the product along with the new features. 3.
Evolution of the product from the previous model

4. Mentioning the positive and negative remarks about the product but in a balanced way. And
also a proper explanation about the judgement made.

5. Comparison between similar products available in the market to give a better review,
especially for the first-time buyers.

6. Audience targeted reviews. This is done to mention which audience group will have most
benefit from this product.

7. Concluding with the summary of the product, whether the product is up to the mark according
to one’s expectations.

11 | P a g e
Social media acts as a huge reservoir of immense amount of data from people all around
the world over a wide range of subjects. Online reviews play a very important role for
the customers to purchase a product go for a movie, go to a restaurant and for
businessman who have to make important business decisions. This data can be very
crucial to understand trend shifts of public opinion as an entity[13][23]. Therefore if this
data undergoes sentiment analysis it can be very helpful for decision making about
anything in the economy. It can be small decisions like choosing amongst two
electronics to something as big as predicting stock market shifts. The reservoir of data
is so huge that labelling it manually may seem like an insane act, therefore systems to
extract or mine sentiments from the reservoir about a certain subject is developed and
later checked for reliability[4]. This not only gives us rich data but helps us get an insight
in to the minds of a large population reviews on a single topic. These are always attached
by star ratings vary from 1-star to 5-star,Which indicates learning about product reviews
of visitors .Figure 1 and figure 2 exhibits an example of a 5-star review on Amazon
website ,which involves Iphone X .

Figure 1: A Product Review on Amazon

However, various types of reviews are spread across various social media website For
example.

12 | P a g e
Figure 2: Is A Tweet on Twitter Which Addresses Negative Review of IPhone
Sentimental Analysis can be used for a wide variety of multi-disciplinary fields. Its basic
use is to understand the basic polarity of sentences and later classify and evaluates them.
There are various ways available on the internet for sentiment analysis. These can be
classified under the supervised ones and the unsupervised ones. Machine learning
languages are used by the supervised ones for their processing. On the other hand, the
unsupervised ones use classified algorithms that work under ruling sets on language
dictionaries. The reservoir of data under consideration has large pools of information
about wide variety of topics. These maybe personal feeling that help read the mind-set
of a huge population. Or it can may have radical characteristics of data about economic
and political situations. Getting access to this information can be of huge benefit while
making decisions. These represent the public mood. These help us reads trends over the
internet in real time therefore have high potential to affect economy in various ways.
Sentiment analysis and natural language processing are probably the most widely
studied upcoming linguistic computation systems in the contemporary research fields.
A system for sematic analysis and evaluation amongst the two is developed and
experimented. A lot of work has been carried out in the field of sentiment analysis
primarily in the field of blogs and product reviews[6]. Work has been carried out on
detecting sentiment in text using algorithms like semantic orientation for detecting
sentiments. Alec Go have tried to develop an algorithm that can that could precisely
classify twitter messages as positive or negative[1][9]. Their primary focus was to achieve
higher accuracy. They have tried to use machine learning techniques in the specific
domain of micro blogs[38][46].

In this Project we are performing a assessment using a suite of evaluation criteria and
data collections using various social media sources. Sentiment analysis is selected so
that the customers get to know which product is good for them according to their needs
by looking into pros and cons[4][6]. By evaluating the results using various aspects and
computer language.

Tokenisation refers to process of splitting a string that are in sequence in to smaller parts
such as words, terms, elements that are called token. Tokens can be keywords or an
entire sentence. In this process, special characters such as punctuation marks are not

13 | P a g e
included. These tokens then act like an input for other processes like mining of text or
parsing. This process is applied in computer science where it plays a crucial role in
lexical analysis.

This process mostly depends on simple shortcuts for separating tokens by the following steps:

• Words or tokens are differentiated among themselves by punctuation mark, white spaces
or line break.

• These spaces between them may or may not be present between them depending on the
degree of need.
• Characters present in any string are parts of token. Tokens may be characters, numbers,
or alpha numeric.

• Tokens sometimes can act as a separator itself. Example, in programming language a

simple summation of two numbers is done by using the plus symbol which although is
a token happens to be a separator.

In a country like India, where it is a bigger consumer than producer in many sectors, the
winning edge among the online shopping experience makes a huge difference. Very
often many companies throw in lots of money to market their brand and forgetting about
the quality of the products. The companies can’t always have a perfect quality control
when it is a mass produced item. The same controlling problems arise in third party
delivery services. So the product company should be thoughtful enough to foresee the
harsh way of dealing with the transportation of the final product from the company to
the consumer’s hand. That’s where companies would make mistake and make the
packaging more aesthetic than strong. We have done an analysis on two popular phones
that have been marking its place in the market in recent years. In an era where one can
get a wide range of phones and as it has become a part of their life, we have chosen a
phone company that is rising year by year and has changed the face of china. The other
company is one of the tech giants for more than two decades with its standards and price
high. We have performed this analysis on the the One plus phone and the I phone. We
have gathered information and used it to make what is called as a word cloud. Graphs
and charts can be helpful when it comes to knowing the company’s profit loss paths and
other business related information. But this won’t work if the data is text based. With a
word cloud however one can get a visualisation of text based data where it can emphasise

14 | P a g e
and highlight the more important word while diminishing the less important ones. This
way, the word that catches your attention first will be the bigger sized word which will
also be the more important one. It works on an algorithm where it sees for the most
repeatedly used word and makes it the most important one. Similarly, the size of other
words depends on how recursive it is. Such comparisons can help in getting to better
conclusion. We have tried to bring an example of how it works using R software which
is one of the most used math software among engineers. On a broad categorisation, there
are 3 types of word clouds based on what it means rather than how it looks. In one type,
there will be a tag for the each item’s subsequent frequency, but in the second type of
word clouds, the frequencies all items and users are aggregated to form a tag on a global
basis. In the third one, this cloud will have categories, with a particular size showing the
number of subcategories it has.

• On the basis of frequency: In this, the size represents how many times that same tag has
been applied to an item. This is comes in handy when we need to display a metadata of
an item which has been democratically elected and where high accurate results are not
required. The second one is the more commonly used type. Here, depending on a tag’s
popularity, a size showing the number of items a tag has been associated with is
acquired.

• On the basis of significance: In this type, the size is used to showcase the importance of
words and their co-occurrences when compared to a corpus rather than frequencies. This
method cannot be used all alone, but it depends on comparing it among the frequencies
in document and expected distributions.

• On the basis of Categorization: In this type, tags are used an approach to categorize the
content items. Tags are constituted in a cloud and there are larger tags that indicate the
quantity of items in that category.

15 | P a g e
3.Literature Review
Table1: Literature Review

S.No Title of the Abstract Significant Outcome

paper

1) Twitter The purpose of their project Use of machine learning

Sentiment was to build an algorithm algorithms to classify user
that can accurately sentiments on
Analysis
classify Twitter twitter to
messages as research public
positive or negative, with opinion about company or a
respect to a query term product.

16 | P a g e
2) Topic In this paper, instead of Our investigation illustrates
Sentiment presenting the sentiment
that three types of information
polarity of each tweet relevant
Analysis in Twitter: A is useful to address the task,
to the topic,we focus our study
Graph-based on hashtag-level sentiment including
classification.
Hashtag (1) sentiment polarity of tweets
Sentiment containing the hashtag (2)
Classification hashtags co-occurrence
Approach relation-ship

(3) the literal meaning of

hashtags.
3) A survey on This paper presents a rigorous This survey covering
opinion mining survey on sentiment analysis, published literature during
and sentiment which portrays views presented 2002-2015, is organized on the
analysis: Tasks, by over one hundred articles basis of sub-tasks to be
approaches and published in the last decade performed, machine learning
applications regarding necessary tasks, and natural language
approaches, and applications of processing techniques used
sentiment analysis. and applications of sentiment
analysis. The paper also

17 | P a g e
presents open issues and along
with a summary table of a
hundred and sixty one articles.

S.No Title of the page Abstract Significant

Outcome

4) A study of Information In this paper, we explore We show that variants of

Retrieval weighting whether more sophisticated the classic tf.idf scheme
schemes for sentiment feature weighting schemes
adapted to sentiment
analysis from Information Retrieval
can enhance classification analysis
accuracy.
provide significant
increases in accuracy,
especially when using a
sublinear function for
term frequency weights

and document
frequency smoothing

18 | P a g e
5) A System for This paper describes a system for
Real-time real-time analysis of public
sentiment analysis can
Twitter sentiment toward presidential
help explore how these
Sentiment candidates in the 2012 U.S.
events affect public
Analysis of 2012 U.S. election as expressed on opinion. While
Presidential Twitter, a microblogging traditional content
service.
Election Cycle analysis takes days or
weeks to complete, the
system demonstrated
here analyzes sentiment
in the entire Twitter
traffic about the election,
delivering

results instantly
and
continuously.

19 | P a g e
6) Sentiment This paper introduces an approach experiments using a
analysis using to sentiment analysis which uses feature set
support vector support vector machines enriched with topic
machines with information on a smaller
(SVMs) to bring together diverse dataset of music reviews
diverse sources of potentially pertinent hand annotated for topic
information information, including several are also reported, the
sources results of which suggest
favourability
that incorporating topic
measures for phrases information into such
and models may also yield
adjectives and, where available, improvement
knowledge of the topic of the text.

S.No. Title of the paper Abstract Significant Outcome

20 | P a g e
7) Sentiment User reviews and They presented a web based
Analysis for comments on hotels on the opinion mining system for hotel
web are an important reviews and user comments that
Hotel
information source in travel supports the hotel management in
Reviews planning. Therefore, monitoring what is published on
knowing about the web about their houses. The
these comments is system is capable of detecting and
important for quality retrieving reviews on the web, to
control to the hotel classify and analyze them, as well
management, too. as to generate comprehensive
overviews of these comments.

8) Sentiment In this work they have used Their results show that projects
analysis of lexical sentiment analysis to developed in Java tend to have
study emotions expressed in more negative commit
commit
commit comments of different comments, and that projects that
comments in open source projects and have more distributed teams tend
GitHub: an analyze their relationship with to have a higher positive polarity
different factors such as used in their emotional content. , they
empirical study programming language, time found that commit comments
and day of the week in which written on Mondays tend to a
the commit was made, team more negative emotion.
distribution and project
approval.

21 | P a g e
9) Sentiment analysis Twitter is a popular micro They used the sentiment
of twitter blogging service that is analysis technique as a valid
used to read and write
audiences: popularity indicator or First,
millions of short messages
they distinguished between
Measuring the on any topic within a 140
character limit. the positive and negative
positive or
Popular or audiences of
negative influence influential users tweet
their status and are popular users.
of popular
retweeted, mentioned, Second, they found
twitterers or replied to by their that the sentiments
audience. Sentiment
expressed in the tweets by
analysis of the tweets by
popular users and their popular users influenced the
audience reveals sentiment of their audience.
whether the audience is Third, from the above two
favourable.
findings they developed
a

positive negative measure

for this influence.

22 | P a g e
S.No Title of paper Abstract Significant Outcome

10) Sentiment Tweets are produced in This paper offers three main
analysis on many languages, but contributions in this
most work on sentiment respect. First, by presenting
Italian tweets
analysis is done for the first of corpus of
English (even tweets for Italian, built
independently of in such a way that makes
Twitter). This is also it possible to use the
due to exact same strategy to build
the availability of similar resources
tools and resources. for other languages
without any manual
intervention.

11) Sentiment They are a wide platform full They developed a framework that
Analysis in of Despite great avails, and makes it possible to analyze
Arabic tweets Twitter comments or “Tweets” as
with the increasingly
having positive, negative or
enormous number of neutral sentiments. This can be
applied in a wide range of
Arabic users on the internet,
applications ranging from
this paper deals with Arabic
politics to marketing. This
Sentiment Analysis.
framework has many novel
aspects such as handling Arabic
dialects, Arabizi and emoticons.
Also, crowd sourcing was
utilized to collect a large dataset
of tweets

23 | P a g e
12) Visual sentiment This paper uses Visual To explore high-volume twitter
sentiment analysis on twitter data, they introduce three novel
analysis on data streams for high time-based visual sentiment
volume twitter analysis techniques: (1)
topicbased sentiment analysis
that

twitter data data which are otherwise extracts, maps, and measures
streams not so easy. customer opinions; (2) stream
analysis that identifies
interesting tweets based on their
density, negativity, and
influence characteristics;

S.No Title of the paper Abstract Significant outcome

13) Survey of mining and In the modern era of social Production of consistent
media (i.e., reviews, forum results with the large
sentiment
discussions, blogs and social amount of data to be
analysis networks), people or processed is a difficulty
company are able to get
hence an automated opinion
public opinions through these
mining and summarization
platforms for their decision
making. .
systems are needed to
overcome subjective biases
and mental limitations with
an objective sentiment
analysis system.

24 | P a g e
14) A Semantic In this paper the difference Common sense
Resource for between the AI system and reasoning is a good solution
for sentiment analysis
Sentiment human intelligence to
but the
Analysis Based on harness common sense scalability of common
Conceptual knowledge obtained from a sense knowledge
Primitives lifetime learning and bases is a major
factor that
experience in informed jeopardizises the
decision making is efficiency of concept
mentioned.

extraction and
Polarity detection.

25 | P a g e
15) Who cares about Sarcasm occurs frequently detection of sarcasm in
sarcastic in user-generated content tweets, while useful, is not
such as blogs, forums and sufficient for accurate
tweets?
micro posts, especially in sentiment analysis of such
Investigating the English, and is inherently tweets. Adding rules to deal
impact of sarcasm on difficult to analyse, not with the scope of sarcastic
sentiment only for a machine but even hashtags does, however,
for a human. One needs to improve performance
analysis have a good understanding considerably, though further
of the context of the improvements could still be
situation, the culture in made.
question, and perhaps the
very specific topic or
people involved in the
sarcastic statement.
Furthermore, tweets
labelled with the hashtag.

S.No Title of the Abstract Significant Outcome

paper

26 | P a g e
16) Sentiment The growing importance of using microblogs and
Analysis and sentiment analysis coincides with twitter they have tried to
the growth of social media such collect the different
Opinion
as reviews, forum discussions, opinion of
Mining blogs, micro-blogs, Twitter, and people about a
social networks. For the first time product and then based
in human history, we now have a on the data they
huge volume of have tried
analyze different
sentiments and

opinionated data recorded in presented in form of pie

digital form for analysis. charts.

17) Recognizing This paper presents a new based on the following

Contextual approach to phrase-level features they tried to
sentiment analysis that first
Polarity in determine the polarity of the
determines whether an
Phrase-Level expression is neutral or polar and review given by the user
Sentiment then disambiguates the polarity of
the polar expressions. With this •Word token
Analysis
approach, the system is able to •Word part of speech •Word
automatically identify the
context
contextual polarity for a large
subset of sentiment expressions, •Prior polarity positive
achieving results that are
negative or neutral
significantly better than baseline.
•Reliability strong subject or
weak subject

27 | P a g e
18) Twitter as a Corpus In our paper, we focus on •We have presented a method
for an automatic collection of a
for using Twitter, the most
corpus that can be used to train
popular microblogging a sentiment classifier. We used
Sentiment
platform, for the task of TreeTagger for POS-tagging
Analysis and and observed the difference in
sentiment analysis. distributions among positive,
Opinion
We show how to negative and neutral sets. From
Mining
the observations we conclude
automatically collect a
that authors use syntactic
corpus for sentiment structures to describe emotions
analysis and or state facts. Some POS-tags
may be strong indicators of
opinion mining purposes. We
emotional text.
perform
linguistic analysis of
the collected corpus and
explain discovered
phenomena.

Using

S.No Title of the Abstract Significant Outcome

paper

28 | P a g e
19) A on Sentiment analysis seeks to they have shown that
Sentimental identify the viewpoint(s) employing the
Education: underlying a text span; an minimum -cut
example application is framework results in
Sentiment
classifying a movie review as the development of
Analysis
efficient algorithms
Using “thumbs up” or “thumbs
down”. To determine this for sentiment
Subjectivity
sentiment polarity, they analysis.
Summarizatio
propose a novel Utilizing
n Based machinelearning method that
contextual
applies text-categorization
Minimum information via this
techniques to just the
Cuts subjective portions of the framework can lead to
document. Extracting these statistically
portions can be implemented
using efficient techniques for significant
finding minimum cuts in improvement in
graphs; this greatly facilitates
polarityclassification
incorporation of cross -
accuracy
sentence contextual
constraints

29 | P a g e
20) Sentiment feature-based sentiment •They then described
analysis featurebased sentiment
Analysis and
analysis which exploits the
Sentiment analysis
Subjectivity , full power of the abstract
of comparative model. After that we discussed
2010
sentences the problem of analyzing
comparative and superlative
3. Opinion search and sentence. Such sentences
retrieval 4. Opinion spam and represent a different type of
evaluation from direct
utility of opinions
opinions which

have been the focus of the

current research.

21) Twitter The purpose of their project Their hypothesis was that
Sentiment was to build an algorithm they could obtain high
that can accurately accuracy on classifying
Analysis
classify Twitter sentiment in Twitter
messages as messages using machine
positive or negative, with learning techniques.
respect to a query term.

S.No Title of the Abstract Significant Outcome

paper

30 | P a g e
22) Analysis in Twitter: A In this paper, instead of Development of a baseline
pre-senting the approach on sentiment analysis
Graph-based
sentiment polarity of each results of the tweets containing
Hashtag
tweet relevant to the topic, the hashtag through simple
Sentiment
they focus their study on voting strategy. In order to
Classification
hashtag- improve the hashtag level
Approach
sentiment classification, use of
level sentiment
classification. a graph model to boost the
results from the voting baseline,
which effectively incorporates

the tweets sentiment

information and hashtags
cooccurrence relationship. The
preliminary results
demonstrate that our graph
model is able

23) A survey on opinion This paper presents a Apart from SVM, NN and
mining rigorous survey on lexicon based approaches; we

31 | P a g e
and sentiment sentiment analysis, found that some of the
analysis: Tasks, which portrays views intelligent techniques have not
approaches and been exploited exhaustively
presented by over one
applications like random forest,
hundred articles evolutionary computation,
published in the last decade association rule mining, fuzzy
regarding necessary rule based systems, rule miner,
conditional random field theory
tasks, approaches, (CRF), formal concept analysis,
and applications radial basis function neural
of network (RBFNN), and online
learning algorithms.
sentiment analysis

24) A study of Most sentiment analysis for accurate classification it is

Information approaches use as baseline a important to use term
Retrieval support vector machines
weighting functions that scale
weighting (SVM) classifier with binary sublinearly in relation to the
unigram weights. number of times a term occurs
schemes for
in a document and that
sentiment In this paper, they explored document frequency smoothing
analysis whether more sophisticated is a significant factor.
feature weighting schemes
from Information Retrieval
can enhance classification
accuracy.

S.No Title of the paper Abstract Significant outcome

32 | P a g e
25) A System for This paper describes a Use of the Twitter
Real-time system for “firehose”and expert-curated
real-time analysis rules and keywords to get a
Twitter
of public full and accurate picture of the
Sentiment sentiment toward online political landscape.
Analysis of 2012 presidential candidates in Use of real-time data
the 2012 U.S. election as processing
expressed on Twitter,

U.S. Presidential a microblogging service. infrastructure and statistical

Election Cycle sentiment model evaluates
public sentiment changes in
response to emerging political
events and news as they
unfold.

26) Sentiment This paper introduces an The method introduced in this

analysis using approach to sentiment analysis paper allows several methods
of assigning semantic values
support vector which uses support
to phrases and words within a
machines with vector machines text to be exploited in a more
useful way than was
diverse (SVMs) to
previously possible, by
information bring together incorporating them as features
sources
diverse sources of for SVM modeling, and for
potentially pertinent explicit topic information to
be utilized, when available,
information, including
by features incorporating such
several favorability values.
measures for phrases and
adjectives and,
where available,

knowledge

33 | P a g e
27) Twitter We introduce a novel We show that using
Sentiment approach for emoticons as noisy labels for
training data is an effective
Classification automatically way to perform distant
using Distant classifying the supervised learning. Machine
Supervision sentiment of learning algorithms (Naive
Twitter messages. Bayes, maximum entropy
These messages classification, and support
are classified as either vector machines) can achieve
positive or negative with high accuracy for classifying
respect to a query sentiment when using this
term. This is useful for method.
consumers who want to
research the sentiment
of

products before
purchase, or companies that
want to monitor the public
sentiment of their brands.

S.No Title of the Abstract Significant Outcome

paper

34 | P a g e
28) Building Recognizing polarity Sentiment analysis is a recent
Lexicon for requires a list of polar attempt to deal with evaluative
words and phrases. For
Sentiment aspects of text. In sentiment
the purpose of building
Analysis such lexicon analysis, one fundamental
from automatically, a lot of problem is to recognize whether
studies have given text expresses positive or
Massive
investigated (semi-)
negative evaluation. Such
Collection of unsupervised method of
learning polarity of words property of text is called polarity
HTML
and phrases. In this paper,
Documents we explore to use
structural clues that can
extract polar sentences

29) Cross- We explore the adaptation of The surge in attention paid to

Linguistic English resources and automated analysis of text
techniques for text sentiment sentiment has largely been
Sentiment
analysis to a new language, focused on English. In this paper,
Analysis: Spanish. Our main focus is the we have discussed how to adapt
modification of an existing an existing English semantic
From
English semantic orientation system to Spanish

35 | P a g e
English to orientation calculator and the while at the same time comparing
Spanish building of dictionaries; several alternative approaches
however we also compare
alternate approaches,
including machine translation
and Support Vector Machine
classification.

30) sentiment This paper uses Visual To explore high-volume twitter

analysis on sentiment analysis on data, they introduce three novel
twitter using time-based visual sentiment
twitter data streams for high
machine analysis techniques: (1)
learning volume topicbased sentiment analysis that
techniques. twitter data extracts, maps, and measures
which are customer opinions; (2) stream
analysis that identifies interesting
otherwise not so easy tweets based on their density,
negativity, and influence
characteristics.

SUMMARY

Sentiment analysis and natural language processing are probably the most widely studied
upcoming linguistic computation systems in the contemporary research fields . A system
for sematic analysis and evaluation amongst the two was developed and experimented [6]
. Sentimental analysis was used in real time during the election campaign of president
Trumph [37] . Amazon Mechanical Turk was used to create baseline sentimental model.
Turkers described their gender , age political orientation. On showing some tweets, they
were asked to understand the sentiment of those tweets (positive, neutral, negative, or
unsure), whether the tweets were funny or sarcastic in nature, the review on a scale from

36 | P a g e
positive to negative, and the tweet author's political view on a scale from old school to
easy [2][10]. Their sentiment model is based on the sentiment label and the sarcasm and
funny labels.

Data Source for this was micro-blogging service Twitter as the data source because it is a
main source of political commentary and discussions[12][15]. All related tweets in real time
were gathered from entire twitter traffic via Gnip Power Track, commercial twitter data
provider . Processing of the tweets was an important task. The text of tweets differs from
the text in articles, books, or even spoken language. As standards of NLP practices, the
text is tokenized for after processing. Using of certain rules to handle the special cases for
the tweets was done. Comparing of several Twitter-specific tokenizers was carried out,
such as TweetMotif and it was found that Christopher Potts’ basic Twitter tokenizer was
found to be appropriate with the base. In conclusion, the tokenizer perfectly handled
common emoticons, URLs, calling numbers, hash tags, HTML tags and twitter mentions,
numbers with decimals and fractions, repetition of Unicode characters & symbols[29][5]. A
System for Sentiment Analysis of Arabic Using Human Computation. The main problem
was finding and analysing informal words from Arabic websites to know the nature of
their meaning

Their source should be informal or colloquial in popular areas. The challenging part here
becomes the in-formalness of the colloquial Arabic used on websites as this may differ
across a range of dialects and geographies.in order to overcome this issue human based
computing is used to deliver the required results[17]. Arabic language got compelled
thought and was so far rejected in terms of the ordinary language dealing with research.
Arabic is a Semitic language and its linguistic structure and lexis is significantly different
[44]
lingos in, express, the Indo-European family (for instance English, German, French,
Hindi .The structure of Arabic words is continuously amazing appeared differently in
relation to English for instance[10][29]. Arabic words are made through a mix of a root (in
any event 3 consonants) with a model, in this manner a rootand-model morphology. In
order to test the practicality of the count system needed the general language corpus
likewise, a phenomenal language corpus[11][48]: their cash related corpus contained 8,815
compositions and 1.48 million tokens circulated from March to August 2005 and posted
by Reuters Arabic organization .

37 | P a g e
4-EXPERIMENTAL PAGE

First step in the experiment performed is collection of data where we scraped data from
amazon website. It contained reviews from the customers on iphone 10 and one plus 6
product delivery. We will be using R code software to perform text mining where
frequently appeared words will be displayed. The tabulated words om computing will
provide a pictorial representation of the result.

38 | P a g e
DATA
COLLECTION

TEXT
MINING

CLOUD
FORMATION
Some of data collected are:-

39 | P a g e
40 | P a g e
41 | P a g e
Figure 3: Spared User Comments

Text mining is done with the help of programming language such as R language. It is
often used for calculating in statistical and graphical situations. This language is widely
used by text or data miners for obtaining a software to analyse data and statistics.

The code used by us for the experiment is mentioned below:

42 | P a g e
Figure 4. R code Used

43 | P a g e
5-RESULTS AND DISCUSSION

Every industry works with a moto of increasing profits as it at the end of the day that
is what pays for the bread and butter of everyone involved in the industry. This is
achieved by when the customers of the company, the ones who are paying and helping
the company generate revenue are satisfied with the goods and services provided to
them. A satisfied customer tends to spend more and show loyalty towards the company.
This satisfaction of customers is very critical to the company. A company may have
various ways to retrieve a customer’s emotions towards the product or service provided
through feedback forms, online reviews, etc[30]. But reading a huge number of online
reviews may seem mundane there for sentiment analysis comes in the picture as a
savour. The algorithms help the industry to scrap reviews from the internet and get a
better idea of the emotions that their customers have towards the company.

Humans have always tried to computerise the works that seems tedious but relevant.
Sentimental analysis has increased the sample space taken for the analysis causing the
result to be more accurate[39]. Unlike opinion mining it has the ability to recognize the
emotion in big sets of texts. It is so effective in many fields like politics, where it shows
the effects of social media opinions[16][49]; journalism, where it can give the propaganda
of a text and marketing, where it can show how the trend is. It can even detect sarcasm
that even most humans find. From an input of raw unstructured text the system
segregates a bunch of needful keywords and then ranks them. The ranks are then
summed up to understand the actual opinion Although this can find a person’s
sentiment about a product or service, it can’t show how it changes with time i.e. it
cannot cope dynamic changes, like the opinions on stocks in stock market which are
unpredictable and rapid. This can be bettered by using lexicon-reduction method. By
this method similar keywords are grouped together. This helps in finding the trend in
dynamic study. However, it was found from a research that sentimental analysis can’t
be used as the only factor for predicting the market [14][19]. Sentimental analysis tools
like NLTK can check 10,000 of them in 3.3 seconds at a precision and recall of 0.56
and 0.49. Whereas for the same sample space, another tool named R sentiment has a
precision of 0.31 and recall of 0.29 and takes 41 minutes. As R sentiment is a newer
tool there might be better results in later versions. For more advanced and accurate
results machine learning can be used which will improve itself from mistakes. There
is still room for improvements in this field[32]. System have coordinated a basic manual
appraisal and found the precision of extraction to stretch out between 60-75%. More
work is required in this one of a kind situation. Sentiment analysis is contextual mining
of text which identifies and extracts subjective information in source material, and
helping a business to understand the social sentiment of their brand, product or service
while monitoring online conversations. However, analysis of social media streams is
usually restricted to just basic sentiment analysis and count based metrics.

44 | P a g e
A customer’s reviews towards a product or service open a vast horizon for
improvement of the good or service as it gives a major insight about the personal
thoughts of customer. So the industry aims to make products that suit their customers
better as this would in turn help them increase their market share as their product will
end up being better to other products available in the market. Once a product achieves
a core competence for being better it the revenue goes up as the sales increase. This
proves that how critical feedback data of customer is to an industry[34]. Sentiment
analysis can be done on a vast variety of application[26]. But in this regard we talk about
it being used to scrap and analyse reviews of products. Particularly talking about
reviews about phones and tablets. With the boom of market now we have a wide variety
of phones and tablets available in stores. They vary in configurations and features and
each product delivers some or the other aspect with arises different sorts of emotions
in customers. Reviews consist of these emotions that help the companies understand
their products market stand.

When a customer buys a product he receives goods and services from a company for
which he has paid good money. Now the customer may have different kinds of
expectations from the product. We classify these expectations into three categories.
The first category is the unspoken basic expectations that the customer has. These
cover expectations that the product should basically cover. For example when we order
food at a restaurant, we expect the food to be served hot. This isn’t something we would
tell the waiter. This is something that is understood without mentioning. Similarly
when we talk about phones, the unspoken expectations are no scratches in the phone,
good packaging, mint quality, etc. when a customer who recently bought a phone
doesn’t receive these traits he may want to express concern against the company
because these are very basic requirements. The second category can be called the
requested expectations. These cover expectations in the product that the customer
specifically asked for or requested for. For example when we order food at a restaurant
and if we ask for extra salt in food or less spicy food, we expect the cook to respect our
request and deliver food with such considerations. Similarly when a customer buys a
phone with certain specifications like it’s RAM or disk size, he expects the phone to
have the exact specifications that have been mentioned. If the phone fails to provide
its specified features this would mean it has failed to deliver what was asked by the
customer. The customer even though maybe partially satisfied with the product but
may still feel there is room for improvement for the company. A company may value
these concerns very highly as these show them room for improvement of their product
and therefore make a superior product. These emotions that have been expressed by
the customer are shown in comments and reviews for the product [27]. The third and the
last type of expectation that a customer may have are surprise or extra expectations.
These expectations cover all the extra things that a service provider may do to impress
the customer and receive loyalty. For example when we order food at a restaurant, a

45 | P a g e
complementary side dish like fries with a burger or a welcome drink are the restaurants
way of impressing the customer. These help the restaurant to gain loyalty from the
customer as these increase the level of satisfaction driver from the food that he paid
for. Similarly in the case of phones this extra expectation is met by gifted accessories
like earphones or phone covers. These give a level of satisfaction to the customer
beyond his expectation therefore making him feel special and therefore starts to rely
on the phone company more. These can also be Antivirus subscriptions or different
apps and features that make a person value the phone beyond his expectations. These
help the company experiment with technologies and promotions that help increase their
products market share. This also helps them know which of these experiments are
fruitful and which ones go under appreciated by the customer. Since these are extra
non-paying gestures by the company the company wouldn’t want to spend its valuable
capital on gestures that do not increase customer satisfaction. Therefore by analysing
the reviews helps the company know about such gestures[7]. Each company tries to
deliver maximum achievable level of satisfaction to its customers for the give costs by
its products. A satisfied is very valuable to a company as each one has a potential to
get more customer or else simply show loyalty towards the company and be back for
future products.

46 | P a g e
Fig 5: word frequency generated
The first result of the study gives us a word frequency of products. This can be visually
seen in Figure 3. So here we see the words that have been repeated the maximum
number of times. This has been done by the system. The system has recognised the
number of times a word has been used and portrayed it in the manner of a list. We see
that for the One plus phone word quality has been used several times and also verified.
This means that the customers seem pleased by the product as they seem to like the
quality of the phone and are satisfied. The attributes like good and verified show that
the customers are satisfied and that the product is good. This also tells the manufacturer
a feedback that his products quality is good and he needs to continue maintaining these
standards for gaining customer loyalty. People talked about the charging and battery
life for this phone. This shows that people are satisfied with the battery life. One plus
introduced higher and bigger batteries in its phones that can be charged very quickly.
Their newer technologies about dash charging seems t have caught on interest with the
customers. This feedback tells the manufacturer that his innovation has helped him
gain competitive advantage over the other brands. This is very important to gain
market leadership. If one plus continues the good work it may have a very bright future.
We see that for the One plus phone word purchase has been used several times and
also verified. This means that the customers seem pleased by the product as they seem
to like the quality of the phone and are satisfied. The attributes like nice and value
show that the customers are satisfied and that the product is good. This also tells the
manufacturer a feedback that his products quality is good and he needs to continue
maintaining these standards for gaining customer loyalty. People talked about the
charging and battery life for this phone. This shows that people are satisfied with the
battery life. One plus introduced higher and bigger cameras in its phones that have
very high resolution. Their newer technologies about pop up cameras seems to have
caught on interest with the customers. This feedback tells the manufacturer that his
innovation has helped him gain competitive advantage over the other brands. This is
very important to gain market leadership. If one plus continues the good work it may
have a very bright future.

On the customers hand is a customer sees that one plus phone’s word frequency he or
she may realise that what the other customers feel about the product and he himself
will be able to make a wiser decision about the phone. We see that for the One plus
phone word quality has been used several times and also verified. This means that the
other users seem pleased by the product as they seem to like the quality of the phone
and are satisfied. The attributes like good and verified show that the other users are
satisfied and that the

product is good. This also tells the new customer that products quality is good and
choosing this one could show fruitful future. Users talked about the charging and

47 | P a g e
battery life for this phone. This shows that people are satisfied with the battery life.
One plus introduced higher and bigger batteries in its phones that can be charged very
quickly. Their newer technologies about dash charging seems to have caught on
interest with all users. This information tells the new customer that the product has
advantage over the other brands. We see that for the One plus phone word purchase
has been used several times and also verified. This means that the customers seem
pleased by the product as they seem to like the quality of the phone and are satisfied.
The attributes like nice and value show that the customers are satisfied and that the
product is good.. People talked about the charging and battery life for this phone. This
shows that people are satisfied with the battery life. One plus introduced higher and
bigger cameras in its phones that have very high resolution. Their newer technologies
about pop up cameras seems to have caught on interest with the customers. This tells
the customer that the product if technically forward and choosing it may be a god
decision.

Now when we talk about the iPhone when we see that word frequencies for the iPhone
we see find various kinds of information available to us. We see that for the iPhone
word expensive has been used several times and also verified. This means that the
customers seem worried that the product is very costly and may not be value for
money. But they are satisfied with the quality as the iPhone is a well optimised product.
The attributes like good and dependable show that the customers are satisfied and that
the product is good. This also tells the manufacturer a feedback that his products
quality is good and he needs to continue maintaining these standards for gaining
customer loyalty. People talked about the display and screen size for this phone. This
shows that people are satisfied with the screen. IPhone introduced higher and bigger
screens in its phones that have high resolutions. Their newer technologies about OLED
seems to have caught on interest with the customers. This feedback tells the
manufacturer that his innovation has helped him gain competitive advantage over the
other brands. This is very important to gain market leadership. If one plus continues
the good work it may have a very bright future. We see that for the iPhone word color
has been used several times. This means that the customers seem pleased by the
product as they seem to like the quality of the phone and are satisfied. The attributes
like grey and silver show that the customers are satisfied and like these colours on the
phone. This also tells the manufacturer a feedback that his products quality is good and
he needs to continue maintaining these standards for gaining customer loyalty. People
talked about the music system for this phone. This shows that people are satisfied with
the music system. IPhone introduced airpods with their new phone and removed the
AUX jack. Their newer technologies about airpods seems to have caught on interest
with the customers. This feedback tells the manufacturer that his innovation has helped
him gain competitive advantage over the other brands. This is very important to gain
market leadership. If one plus continues the good work it may have a very bright future.

48 | P a g e
On the customers hand is a customer sees that one plus phone’s word frequency he or
she may realise that what the other customers feel about the product and he himself
will be able to make a wiser decision about the phone. We see that for the iPhone word
expensive has been used several times and also verified. This means that the previous
users seem worried that the product is very costly and may not be value for money.
But they are satisfied with the quality as the iPhone is a well optimised product. The
attributes like good and dependable show that the users are satisfied and that the
product is good. This also tells the new customer that the product is dependable and
buying it seems like a good idea. People talked about the display and screen size for
this phone. This shows that people are satisfied with the screen. IPhone introduced
higher and bigger screens in its phones that have high resolutions. Their newer
technologies about OLED seems to have caught on interest with all the users. This
information tells the customer that the product seems good as compared to the other
products. This is very important to gain market leadership. If apple continues the good
work it may have a very bright future. We see that for the iPhone word colour has been
used several times. This means that the customers seem pleased by the product as they
seem to like the quality of the phone and are satisfied. The attributes like grey and
silver show that the customers are satisfied and like these colours on the phone. This
tells the new user that most people opt for Grey and Silver colours. So if he wants his
product to look like the others he can also choose one of these colours. But if he wants
a certain degree of uniqueness in the product he can opt for other colours available.
People talked about the music system for this phone. This shows that people are
satisfied with the music system. IPhone introduced airpods with their new phone and
removed the AUX jack. Their newer technologies about airpods seems to have caught
on interest with the customers. This information tells the new customer that buying
this product seems like a good idea as they keep experimenting with new technologies.
This way the system helps both the manufacturers and even the users to make better
decisions when it comes to these phones.

Fig: 6- word cloud from IPhone and One plus

So clearly from the word map in figure 4 we will see what the shoppers emotions
tell us concerning their demand and expectations being met by the merchandise. we
have a tendency to see that almost all folks comment concerning the colors of the

49 | P a g e
merchandise. The phone comes has 3 variants particularly house gray, silver and
golden. folks have largely {preferred|most popular|most we have a
tendency
the house gray variant so we see that this is often a most well-
liked selection therefore the manufacture shouldn't discontinue this color choice in
later models. equally folks have talked concerning the screen. The screen
is

is
manufactures to create some enhancements. one amongst the
main issues that's shown by the word cloud is that the merchandise is dear. A person’s
utility is additionally variable to the price he has purchased the
merchandise. therefore a pricey product ought to drive loads higher
utility however this doesn’t appear to be the case for the phone.
The show has continually been believed to be higher on iPhone however these
complaints concerning the camera within the phone tells the manufactures to

50 | P a g e
has continually been believed to be higher on iPhone however these
complaints concerning the camera within the phone tells the manufactures to
create some enhancements. one amongst the main issues that's shown by the word
cloud is that the merchandise is dear. A person’s utility is additionally variable to the
price he has purchased the merchandise. therefore a pricey product ought
to drive loads higher utility however this doesn’t appear to be the case for the
phone. so the corporate ought to explore for ways that to scale back the prices of
its product for its users. Comments may also be seen concerning the look and weight
of the merchandise. When we explore the word cloud for one and phone.
There square measure several things that may be inferred. individuals have
raised considerations regarding the screen size being to massive. It goes on the far
side the comfort of the palm of an individual. the corporate ought to work
o up the bioengineering of the phone. individuals compliment the camera of the
phone because it is seen in such a large amount of photos. The phone
termed worth for cash because it could be a fraction of the price of the opposite phone
in thought. because the prices square measure low the client satisfactions square
measure high. individuals additionally refer the face unlock and the way it cases
errors within the dark places this will be terribly painful for the user and a game
changer once it involves a person’s loyalty to the corporate. a corporation can’t afford
to lose its loyal customers because it reduces revenues and also the general market
opinion of the corporate. therefore the company ought to see to that to correct these
errors in order that the client stay glad. individuals additionally refer the speaker and
the way it cases errors within the thronged places this will be terribly painful for the
user and a game changer once it involves a person’s loyalty to the corporate.
corporation can’t afford to lose its loyal customers because it reduces revenues and
also the general market opinion of the corporate[18][47]. therefore the company ought
to see to that to correct these errors in order that the client stay glad. individuals have
raised considerations regarding the show being dark to be used in daylight. It goes
on
the far side the comfort of human eye. the corporate ought to work
o up the bioengineering of the phone. individuals compliment the charging speed of
the phone because it is seen in such a large amount of photos. the
corporate provides totally different innovation for the speed of charging as they
saw that individuals need this. This currently has become one in all the
core competence for the corporate. equally individuals have talked regarding the
camera. The bluetooth has forever been believed to be higher on
Oneplus however these complaints regarding the Bluetooth technology within
the phone tells the manufactures to form some enhancements. one in all the most
important considerations that's shown by the word cloud is that the merchandise
dear. A person’s utility is additionally variable to the price he has bought the
merchandise. These inferences facilitate the corporate apprehend what
{the clients|the purchasers|the shoppers} feel regarding the merchandise in
one look and so they will work on the short comings and improve the
merchandise for the next customer satisfaction.The IPhone
incredibly tastily created and is incredibly lightweight weight. this suggests that the
shoppers square measure proud of the actual fact that the merchandise is providing

51 | P a g e
these guarantees. folks even bring up the improvement of the merchandise. The
phone doesn’t typically lag and guarantees a awfully long life while
not abundant troubles over the years. . equally folks have talked concerning the
camera. The camera has continually been believed to be higher on
iPhone however these complaints concerning the camera within the phone tells the
create some enhancements. The speaker has continually been believed to be higher
on
iPhone however these complaints concerning the camera within the phone tells the
manufactures to create some enhancements. These inferences facilitate the
corporate understand what clients|the purchasers|the shoppers} feel concerning the
merchandise in one look and so they'll work on the short comings and improve the
merchandise for a better customer
satisfaction.
Our work is titled as sentimental analysis of
iphone and one plus product delivery. From positive reviews received for one plus
the results obtained, it was found that one are more in number. The word good
plus is better than iphone as the number of
was more frequent in the one plus review. From the results obtained, it absolutely
was found that one and is best than iphone because the range of positive reviews
received for one and area unit additional in range. The
word smart was additional frequent within the one and review. We have a tendency
to were additionally ready to determine that almost all of the issues incurred by
iphone were thanks to miss management within the provide chain network.
Customers had negative response towards product injury and defective accessories.
In
keeping with customers compared with iphone one and had higher reviews owing
to the impeccable battery life and no complaints over broken product delivery[7]. We
were also able to identify that most of the problems incurred by iphone were due
to miss management in the supply chain network. Customers had negative
in
response towards product damage and defective accessories. According to
customers comparison with iphone one plus had better reviews because of the
impeccable battery life and no complaints over damaged product delivery

Now let’s talk about the advantages from a customers end. Here the study of sentient
analysis works for a customer is quite opposite too the company’s way. When we
talked about the company the word cloud gave the company a feedback to help improve
their product. Now when we talk about the customers end we are going see how this
can be used as a decision making tool. When a client buys a product he receives
merchandise and services from an organization that he has paid sensible cash. currently
the client might have completely different forms of expectations from the merchandise.

52 | P a g e
we tend to classify these expectations into 3 classes. the primary class is that the
unspoken basic expectations that the client has. These cowl expectations that the
merchandise ought to primarily cowl. for instance once we order food at a building, we
tend to expect the food to be served hot. This isn’t one thing we'd tell the waiter. this
is often one thing that's understood while not mentioning. equally once we quote
phones, the unspoken expectations are not any scratches within the phone, sensible
packaging, mint quality, etc. once a client World Health Organization recently bought
a phone doesn’t receive these traits he might want to precise concern against the
corporate as a result of these ar terribly basic needs. The second class is known as the
requested expectations. These cowl expectations within the product that the client
specifically asked for or requested for. for instance {when we tend to |once we| after
we} order food at a building and if we provoke further salt in food or less spicy food,
we tend to expect the cook to respect our request and deliver food with such issues.
equally once a client buys a phone with bound specifications like it’s RAM or disk
size, he expects the phone to possess the precise specifications that are mentioned. If
the phone fails to produce its nominal options this could mean it's did not deliver what
was asked by the client. The client even if perhaps partly glad with the merchandise
however should feel there's space for improvement for the corporate. an organization
might worth these considerations terribly extremely as these show them space for
improvement of their product and so create a superior product. These emotions that are
expressed by the client ar shown in comments and reviews for the merchandise [27][44].
The third and therefore the last style of expectation that a client might have ar surprise
or further expectations. These expectations cowl all the additional things that a service
supplier might do to impress the client and receive loyalty. for instance once we order
food at a building, a complementary dish like fries with a burger or a welcome drink
ar the restaurants means of impressing the client. These facilitate the building to realize
loyalty from the client as these increase the extent of satisfaction driver from the food
that he got. equally within the case of phones this further expectation is met by talented
accessories like earphones or phone covers. These provides a level of satisfaction to
the client on the far side his expectation so creating him feel special and so starts to
have confidence the telephone company additional. These can even be Antivirus
subscriptions or completely different apps and options that create an individual worth
the phone on the far side his expectations. These facilitate the corporate experiment
with technologies and promotions that facilitate increase their product market share.
This conjointly helps them apprehend that of those experiments ar fruitful and which
of them go underneath appreciated by the client. Since these ar further non-paying
gestures by the corporate the corporate wouldn’t need to pay its valuable capital on
gestures that don't increase client satisfaction. so by analysing the reviews helps the
corporate understand such gestures. every company tries to deliver most doable level
of satisfaction to its customers for the provide prices by its product. A glad is extremely
valuable to an organization as every one contains a potential to induce additional client
as an alternative merely show loyalty towards the corporate and be back for future
product. This also shows that customer wants to be pleased to the maximum level and
therefore wants to make a wise decision for what he buys after all he is the one who
paid money and would be using the product. He would want the product to atleast last
a decent amount of time and not give any troubles whereas also think about the extra
features and design perfections that he gets with the product.

53 | P a g e
Fig: 7- word cloud from Iphone and One plus

So talking about the application on customers end. A customer can use the word cloud
like the one in figure 5 generated by sentimental analysis of comments and reviews by
the pervious customers to make a wise decision about the product [36]. The customer
wants to be pleased to the maximum level and therefore wants to make a wise decision
for what he buys after all he is the one who paid money and would be using the product.
He would want the product to atleast last a decent amount of time and not give any
troubles whereas also think about the extra features and design perfections that he gets
with the product. On looking at the word cloud for apple iPhone we can see that the
product has space grey highlighted. This means most customers liked the space grey
colour therefore buying that could be a good decision if the customer wants to go with
the public opinion or else he can buy some other colour to get some uniqueness towards
his product[40][44]. So clearly from the word map we can see what the customers
emotions tell us about their requirement and expectations being met by the product. In
figure 16 we see that most people comment about the colours of the product. The phone
comes has three variants namely space grey, silver and golden. People have mostly
preferred the space grey variant therefore we see that this is a preferred choice so the
customer can easily decide his preference. Similarly people have talked about the
screen. The screen has always been believed to be better on iPhone but these
complaints about the camera in the phone tells the new customer to make some a smart
decsion. One of the major concerns that is shown by the word cloud is that the product
is expensive. A person’s utility is also variable to the cost he has paid for the product.
So a costly product should drive a lot higher utility but this doesn’t seem to be the case
for the phone. Therefore the customer should be very critically concerned about his
budget for the phone. Comments can also be seen about the design and weight of the
product. The IPhone is very tastefully made and is very light weight. This means that
most of the users are happy with the fact that the product is providing these promises.
People even talk about the optimization of the product. The phone doesn’t usually lag
and promises a very long life without much troubles over the years. . Similarly people
have talked about the camera. The camera has always been believed to be better on
iPhone but these complaints about the camera in the phone tells the customer to be
concerned about the camera. One of the major concerns that is shown by the word
cloud is that the product is expensive. A person’s utility is also variable to the cost he
has paid for the product. So a costly product should drive a lot higher utility but this
doesn’t seem to be the case for the phone. The display has always been believed to be
better on iPhone but these complaints about the camera in the phone tells the customers
to be concerned about the display. The speaker has always been believed to be better
on iPhone but these complaints about the camera in the phone tells the customer to
seek other options if he really wants good ones on his phone. These inferences help the
customer know what the other users feel about the product in one glance and therefore
they can work on a smart decision that would drive a higher level of satisfaction.

When we look at the word cloud for one plus phone. There are many things that can
be inferred. People have raised concerns about the screen size being too big. It goes
beyond the comfort of the palm of a person. The customer should know that the
ergonomics of the phone aren’t as comfortable. People compliment the camera of the
phone as it is seen in so many pictures. The phone is called value for money as it is a

54 | P a g e
fraction of the cost of the other phone in consideration. This shows that the customer
needs to think about the aspects shown and make a wise decision and he is the one
buying the products and will be the end user. As the costs are low the customer
satisfactions are high. People also talk about the face unlock and how it cases errors in
the dark places this can be very painful for the user and a game changer when it comes
to a person’s comfort of using the phone. People also talk about the speaker and how
it cases errors in the crowded places this can be very painful for the user and a game
changer when it comes to a person’s ease of listening to music. This shows that the
customer needs to think about the aspects shown and make a wise decision and he is
the one buying the products and will be the end user. People have raised concerns about
the display being dark for use in sunlight. It goes beyond the comfort of human eye.
The customer should understand the short coming of the phone. This shows that the
customer needs to think about the aspects shown and make a wise decision and he is
the one buying the products and will be the end user. People compliment the charging
speed of the phone as it is seen in so many pictures. The user like and appreciate
innovation for the speed of charging as they saw that people require this. This now has
become one of the core competency for the company. Similarly people have talked
about the camera. The bluetooth has always been believed to be better on Oneplus but
these complaints about the Bluetooth technology in the phone tells the customer to see
to the usefulness. One of the major concerns that is shown by the word cloud is that
the product is expensive. A person’s utility is also variable to the cost he has paid for
the product. These inferences help the company know what the customers feel about
the product in one glance and therefore they can work on the short comings and
improve the product for a higher customer satisfaction. This shows that the customer
needs to think about the aspects shown and make a wise decision and he is the one
buying the products and will be the end user.

DISCUSSION

A client’s reviews towards a product or service open a huge horizon for improvement
of the great or service because it provides a significant insight regarding the private
thoughts of customer. That the trade aims to create product that suit their customers
higher as this could successively facilitate them increase their market share as their
product can find yourself being higher to alternative product obtainable within the
market. Once a product achieves a core competency for being higher it the revenue
goes up because the sales increase. This proves that however vital feedback knowledge
of client is to Associate in nursing trade. Sentiment analysis will be done on a huge
style of application. However during this regard we tend to observe it being employed
to scrap and analyse reviews of product. Significantly talking regarding reviews
regarding phones and tablets. With the boom of market currently we've got a large style
of phones and tablets obtainable in stores. They vary in configurations and options and
every product delivers some or the opposite side with arises different types of emotions
in customers.

55 | P a g e
Reviews incorporates these emotions that facilitate the sinesses perceive their product
market stand. Every business works with a moto of skyrocketing profits because it at
the tip of the day that's what pays for the bread and butter of everybody concerned
withinthe business[44]. This can be achieved by once the shoppers of the corporate,
those WHO square measure paying and serving to the corporate generate revenue
square measure happy with the products and services provided to them. A happy client
tends to pay a lot of and show loyalty towards the corporate. This satisfaction of
consumers is extremely vital to the corporate. An organization might have numerous
ways in which to retrieve a customer’s emotions towards the merchandise or service
provided through feedback forms, on-line reviews, etc. however reading a large range
of on-line reviews could appear mundane there for sentiment analysis comes within
the image as a savour.

The algorithms facilitate the business to scrap reviews from the net and obtain a more
robust plan of the emotions that their customers have towards the corporate. When a
client buys a product he receives merchandise and services from a corporation that he
has paid smart cash. currently the client might have totally different sorts of
expectations from the merchandise. we have a tendency to classify these expectations
into 3 classes. the primary class is that the unspoken basic expectations that the client
has.these ese cowl expectations that the merchandise ought to essentially cowl. for
instance after we order food at a edifice, we have a tendency to expect the food to be
served hot. This isn’t one thing we'd tell the waiter. this can be one thing that's
understood while not mentioning. equally after we quote phones, the unspoken
expectations aren't any scratches within the phone, smart packaging, mint quality, etc.
once a client UN agency recently bought a phone doesn’t receive these traits he might
want to specific concern against the corporate as a result of these square measure
terribly basic necessities. The second class are often referred to as the requested
expectations. These cowl expectations within the product that the client specifically
asked for or requested for. for instance {when we have a tendency to|once we|after we}
order food at a edifice and if we arouse further salt in food or less spicy food, we have
a tendency to expect the cook to respect our request and deliver food with such
concerns. equally once a client buys a phone with sure specifications like it’s RAM or
disk size, he expects the phone to possess the precise specifications that are mentioned.
If the phone fails to produce its mere options this might mean it's didn't deliver what
was asked by the client. The client although perhaps partly glad with the merchandise
however should still feel there's space for improvement for the corporate. a corporation
might worth these considerations terribly extremely as these show them space for
improvement of their product and so build a superior product. These emotions that are
expressed by the client square measure shown in comments and reviews for the
merchandise. The third and therefore the last sort of expectation that a client might
have square measure surprise or further expectations. These expectations cowl all the
additional things that a service supplier might do to impress the client and receive
loyalty. for instance after we order food at a edifice, a complementary dish like fries

56 | P a g e
with a burger or a welcome drink square measure the restaurants manner of impressing
the client. These facilitate the edifice to achieve loyalty from the client as these increase
the amount of satisfaction driver from the food that he procured. equally within the
case of phones this further expectation is met by talented accessories like earphones or
phone covers. These provides a level of satisfaction to the client on the far side his
expectation so creating him feel special and so starts to think about the public-service
corporation a lot of. These may be Antivirus subscriptions or totally different apps and
options that build someone worth the phone on the far side his expectations. These
facilitate the corporate experiment with technologies and promotions that facilitate
increase their merchandise market share. This additionally helps them recognize that
of those experiments square measure fruitful and which of them go underneath
appreciated by the client. Since these square measure further non-paying gestures by
the corporate the corporate wouldn’t need to pay its valuable capital on gestures that
don't increase client satisfaction. so by analysing the reviews helps the corporate realize
such gestures. every company tries to deliver most realizable level of satisfaction to its
customers for the offer prices by its merchandise. A glad is incredibly valuable to a
corporation as every one incorporates a potential to induce a lot of client as an
alternative merely show loyalty towards the corporate and be back for future
merchandise.

When we talk about Sentiment Analysis as a system, one of the basic problems that the
system faces are the different forms of colloquial language used in the data. The system
being human made cannot detect emotions of the reviewer therefore also fails to
recognise sarcastic comments and slangs present in comments[20][31]. Jokes and sarcasm
are a major part of communication and are so widely used that the system needs some
adaptability to them. They are so common that BBC has a specific site for people who
don’t speak English language to learn the use of sarcasm in the English. The oxford
dictionary states that jokes are sharp or cutting remarks or unsavoury jeer. These days
it is regularly used when people "state something as funny or downright obvious, or
something that opposes to their genuine feelings in a way to be sharp or to reach at an
important resolution", as described on the BBC joke site page referenced already .
(Bousfield, 2007) delineates it as "the use of frameworks that, externally have all of
the reserves that require being reasonable to the situation, also are planned to be taken
as which means the opposite with regard to be taken upto the officials. So the system
loses the ability to recognise such abnormalities. We often come across jokes or slangs
written by reviewers on websites and even in documents. These statements may carry
sentiment defining adjectives that the system is supposed to recognise. Once the system
recognises the sentiments in the sarcastic statements it groups the sentiment as per its
polarity. What the system fails to recognise is the partial sarcasm intended by the writer
which turns the polarity of the statement. Once grouped in the wrong polarity this
would result in the wrong sentiment scores while evaluating. A similar problem arises

57 | P a g e
when the data consists of slangs from English language or any of the locally spoken
language[51]. Many slangs define sentiments and also many slangs are used as both
positive and negative sentiments. So therefore these cannot be included under the
sorting dictionaries as their polarity is not defines clearly. So either the system chooses
to ignore the slang sentiment or it may have a high possibility to put it into the wrong
polarity group. In all accordance to the Data of public opinion it fails its sole purpose
of mining the right sentiment expressed in the statement [47]. This reflects in the
sentiment scores therefore reducing the reliability of the system results. With the boom
of social media Hashtags are of the most commonly used symbols followed by a wide
variety of emoticons. The system can be loaded with dictionaries that read the emoticon
and know their polarity as positive, negative or neutral. Thus the system can accurately
access emoticons and this ability increases the reliability of the results achieved. But
when we talk about the hashtag symbols we might be looking at another major issue
for the system. The hashtag symbol is used along with a word that combines define a
popular sentiment that the reader should be able to understand. They reduce the need
to specify in detail the subject or keywords of the statement. So these statements also
are of high value as they carry major sentiments. Now that they are so popularly used
they form an essential reading. But as of the technologies and limitations of artificial
intelligence the system is unable to pinpoint the emotion or action that the hashtags
define. This lack of reading causes the sentiment that was defined to be unreadable
therefore opted out of the data. Thus reducing the reliability of results given by the
system. A similar case if faced when the data under consideration has catchphrases or
pop culture slogans or dialogues. These may define an emotion that cannot be read by
the system thus causing similar issues. But to overcome all these issues special case
zones are defined to the sorting dictionaries as zones. Because the human coder has
information of the subject on which the system is supposed to run these special case
zones can be customised and loaded. This would give the system the ability to form
and relate such scenarios to its custom additions and therefore record the sentiment.
Once recorded the sentiment would give its contribution to the sentiment score. The
zones may be referred to as constrained state models as they need to be changed with
every change is subject or topic switch. System thus itemises the decision of the
sentiment that was recorded from the statement in data. To derive unjust insights, it's
vital to grasp what facet of the complete may be a user discussing regarding. For
example: Amazon would need to segregate messages that connected to: late deliveries,
request problems, promotion connected queries, product reviews etc. On the opposite
hand, Starbucks would need to classify messages supported whether or not they relate
to employees behavior, new occasional flavors, hygiene feedback, on-line orders, store
name and site etc. however however will one do that?
We introduce AN intelligent good search algorithmic rule known as discourse
linguistics Search (a.k.a. CSS). The manner CSS works is that it takes thousands of
messages and conception|an idea|a thought|a plan|an inspiration} (like

58 | P a g e
Price) as input and filters all the messages that closely match with the given concept.
The graphic shown below demonstrates however CSS represents a significant
improvement over existing ways utilized by the business. A conventional approach for
filtering all value connected messages is to try to to a keyword search on value and
different closely connected words like (pricing, charge, $, paid). This methodology but
isn't terribly effective because it is nearly not possible to consider all the relevant
keywords and their variants that represent a specific construct. CSS on the opposite
hand simply takes the name of the construct (Price) as input and filters all the
contextually similar even wherever the plain variants of the construct keyword aren't
mentioned.

So we know sentiment analysis is multidisciplinary[14][19]. Therefore we can very easily

use these multidisciplinary systems to deal with the semantics that have been written
by people or customers to achieve a better understanding of the required subject. This
system helps us to record and archive or analyse the statements that we have recorded
as negative, positive, or neutral. we all know sentiment analysis is multidisciplinary.
so we will terribly simply use these multidisciplinary systems to traumatize the
linguistics that are written by individuals or customers to realize an improved
understanding of the desired subject. this technique helps U.S. to record and archive or
analyse the statements that we've got recorded as negative, positive, or neutral. even if
this method looks terribly straightforward and may be done simply by a person's,
however we tend to area unit talking regarding thousands of such statements that will
appear terribly mundane for a person's even be terribly time overwhelming. currently
the issue that arises that these actions want access to varied differing types of
technological dictionaries. These embody the emotional and verbal linguistics.
Whereas additionally bring up the technical aspects of scarping statements of the net
that require to be analysed. Even though this process seems very simple and can be
done easily by a human, but we are talking about thousands of such statements that
may seem very mundane for a human also be very time consuming. Now the thing that
arises that these actions need access to various different types of technological
dictionaries. These include the emotional and verbal linguistics. Whereas also talk
about the technical aspects of scarping statements of the internet that need to be
analysed.

Now while we are already talking about all the various achievability of computers and
how sentiment analysis have found a place in the systems. We need to put some light
on how the inter dependence of humans and computers have risen. We don’t only see
the need or growth of this interdependence but also respect the reliability of the two..
whereas we have a tendency to area unit already talking regarding all the varied
attainableness of computers and the way sentiment analysis have found an area within
the systems. we'd like to place some lightweight on however the inhume dependence
of humans and computers have up. we have a tendency to don’t solely see the necessity

59 | P a g e
or growth of this reciprocality however conjointly respect the reliableness of the 2.
Humans and computers along have sped up the method of mundane activities. similar
to reading comments and realising if they're helpful or not is incredibly necessary for
the employment of some or the opposite analysis. The human is simply answerable for
watching the activities that area unit being done by the pc. This helps icrease the
productivity of each because the omputer will nnow scale back errors at a larger speed
thus the|and also the} human is absolve to be watching alternative sh activities
therefore reducing man power. Humans and computers together have speeded up the
process of mundane activities. Just like reading comments and realising if they are
useful or not is very important for the use of some or the other analysis. The human is
just responsible for monitoring the activities that are being done by the computer. This
helps icrease the productivity of both as the omputer can nnow reduce errors at a greater
speed and the human is free to be monitoring other sh activities therefore reducing man
power

Humans have the power of analysing and solving problem due the help of their skill
and critical thinking but they may lack the speed. Computers on the other hand have
the speed to cover tedious activities at a high speed but lack the ability to judge the
right complexity of the process there can’t solve. . Humans have the facility of
analysing and resolution drawback due the assistance of their ability and demanding
thinking however they'll lack the speed. Computers on the opposite hand have the
speed to hide tedious activities at a high speed however lack the flexibility to evaluate
the proper complexness of the method there can’t solve. Sentiment analysis brings
along the flexibility of humans and computers with the hip of varius dictionaries that
area unit loaded to the system that facilitate it undrerstand and solve issues at a quicker
pace. This method is incredibly wide used all across the industries and package
developers because it is reasonable, effective and extremely user friendly. Sentiment
analysis brings together the ability of humans and computers with the hep of varius
dictionaries that are loaded to the system that help it undrerstand and solve problems
at a faster pace. This process is very widely used all across the industries and software
developers as it is cheap, effective and very user friendly.

So we realised that there are many ways in which one can scrap statements of the
internet and most of them are in English so there are simply usable for us. These
statements carry emotions of the writer there carry sentiments that maybe positive or
negative and using these statements to our benefit for analysis is why we chose
sentiment analysis as an applicable tool to the product marketing and analysing
strategies[19][33]. One can say that sentiment analysis can be used by two ways. The first
one is called supervised as the name suggests it is supervised by various set of rules
that help govern its functioning. we tend to completed that there area unit many ways

60 | P a g e
within which one will scrap statements of the net and most of them area unit in English
therefore there area unit merely U.S.able for us[35]. These statements carry emotions of
the author there carry sentiments that perhaps positive or negative associate degreed
exploitation these statements to our profit for analysis is why we tend to selected
sentiment analysis as an applicable tool to the merchandise selling and analysing
methods. One will say that sentiment analysis may be utilized by 2 ways in which. the
primary one is named supervised because the name suggests it's supervised by varied
set of rules that facilitate govern its functioning[26][42]. This may consist of algorithms
from machine learning and other such technologies that help the system work on a
required path with feedbacks to whether the system in moving in the right
direction[35]de. The other one is called the unsupervised one as it may use dictionaries
that aren’t based on technological strategies but one rules made by humans regarding
speech and colloquial usage.

A hard challenge is faced when we talk about comments from commerce websites.
These websites have comments and reviews about products that are marketed on the
website. These comments hold sentiments of the customers that help the company
know about its benefits and set backs of its products. What the person likes or dislikes
and everything of those sorts. But the problem that arises here is that the language on
these websites is often informal. This shall be discussed in detail in the later paragraph.
Sentiment Analysis as a system, one in all the essential issues that the system faces ar
the various sorts of conversational language utilized in the information [24].
The system being human created cannot notice emotions of the reviewer so conjointly
fails to recognise mordant comments and slangs gift in comments. Jokes and humor ar
a serious a part of communication and ar therefore wide used that the system desires
some ability to them. they're therefore common that BBC encompasses a specific
website for people that don’t speak English to find out the utilization of humor within
the English. The oxford wordbook states that jokes ar sharp or cutting remarks or
unsavoury jeer. We often come across jokes or slangs written by reviewers on websites
and even in documents. These statements may carry sentiment defining adjectives that
the system is supposed to recognise. So the system loses the flexibility to recognise
such abnormalities. we regularly come upon jokes or slangs written by reviewers on
websites and even in documents. These statements might carry sentiment process
adjectives that the system is meant to recognise. Once the system recognises the
feelings within the mordant statements it teams the sentiment as per its polarity. What
the system fails to recognise is that the partial humour supposed by the writer that turns
the polarity of the statement. Once sorted within the wrong polarity this would
lead to the incorrect sentiment scores whereas evaluating. an identical downside arises
once the information consists of slangs from West
Germanic language or any of
the regionally language. several slangs outline sentiments
and additionally several slangs ar used as each positive and negative sentiments.
therefore thus these can't be enclosed below the sorting dictionaries as their polarity

61 | P a g e
isn't defines clearly. therefore either the system chooses to ignore the slang sentiment
or it should have a high risk to place it into the incorrect polarity cluster. all told
accordance to the information of belief it fails its sole purpose of mining the proper
sentiment expressed within the statement. This reflects within the sentiment scores thus
reducing the responsibleness of the system results. With the boom of social media
Hashtags ar of the foremost ordinarily used symbols followed by a large sort of
emoticons. The system is loaded with dictionaries that browse the facial gesture and
apprehend their polarity as positive, negative or neutral. so the system will accurately
access emoticons and this ability will increase the responsibleness of the results
achieved. however once we quote the hashtag symbols we would be gazing another
major issue for the system. The hashtag image is employed at the side of a word that
mixes outline a wellliked sentiment that the reader ought to be able to perceive. They
scale back the necessity to specify well the topic or keywords of the statement.
therefore these statements are of high price as they carry major sentiments. currently
that they're therefore popularly used they type an important reading. however as of the
technologies and limitations of computing the system is unable to pinpoint the feeling
or action that the hashtags outline[21]. This lack of reading causes the sentiment that
was outlined to be illegible thus opted out of the information. so reducing the
responsibleness of results given by the system. an identical case if two-faced once the
information into consideration has catchphrases or popular culture slogans or
dialogues. These might outline associate feeling that can't be browse by the system so
inflicting similar problems. however to beat of these problems special case zones ar
outlined to the sorting dictionaries as zones. as a result of the human engineer has info
of the topic on that the system is meant to run these special case zones is custom-made
and loaded. this is able to offer the system the flexibility to make and relate such
eventualities to its custom additions and thus record the sentiment. Once recorded the
sentiment would offer its contribution to the sentiment score. The zones could also be
remarked as unnatural state models as they have to be modified with each modification
is subject or topic switch. System so itemises the choice of the sentiment that was
recorded from the statement in information. Now that they are so popularly used they
form an essential reading. But as of the technologies and limitations of artificial
intelligence the system is unable to pinpoint the emotion or action that the hashtags
define[21][22]. This lack of reading causes the sentiment that was defined to be
unreadable therefore opted out of the data. Thus reducing the reliability of results given
by the system. A similar case if faced when the data under consideration has
catchphrases or pop culture slogans or dialogues. These may define an emotion that
cannot be read by the system thus causing similar issues. But to overcome all these
issues special case zones are defined to the sorting dictionaries as zones. Because the
human coder has information of the subject on which the system is supposed to run
these special case zones can be customised and loaded. This would give the system the
ability to form and relate such scenarios to its custom additions and therefore record
the sentiment. Once recorded the sentiment would give its contribution to the sentiment
score[36][37]. The zones may be referred to as constrained state models as they need to

62 | P a g e
be changed with every change is subject or topic switch. System thus itemises the
decision of the sentiment that was recorded from the statement in data.

Scraping data from different sources has been done for a long time now as we need to
access information from one or the other source or some or the other use. Now is one
simply sits and copies the data is one way of completing the act but we don’t usually
prefer that as it is mundane and very time consuming.
Scraping information from totally different supplys has been in hot water an extended
time currently as we'd like to access data from one or the opposite source or some or
the opposite use. now could be one merely sits and copies the info is a way of finishing
the act however we have a tendency to don’t sometimes like that because it is mundane
and really time intense. thus mistreatment technologies to finish such mundane
procedures was adopted at a really early stage. Sentiment analysis is one amongst these
ways that of overcoming the time consumption problems with scraping activities. . And
leverage their business. firms nowadays complete that they need the flexibility to act
as a catalyst that helps every one deliver higher business resolutions to the prevailing
aspects of the issues visaged. Therefore using technologies to complete such mundane
procedures was adopted at a very early stage. Sentiment analysis is one of these ways
of overcoming the time consumption issues of scraping activities. . And leverage their
business. Companies today realised that they have the ability to act as a catalyst that
helps each one deliver better business resolutions to the prevailing aspects of the
problems faced. The data that we receive from the internet may be useful or completely
opposite as we never know what to be trusted. But let’s just agree on the possibilities
of various users and out of them most tend to be reliable therefore the scrapped data
even seems to be reliable.

Everything that has been discovered or invented has come with its own sets of merits and
demerits. We ourselves are never able to stop ourselves to only use something for its
benefits. Sooner or later we tend to deviate from the unrealistic expectations of idealistic
scenarios and move towards other possibilities. Same can be said about data scraping. It
is an amazing technology with lots of merits and demerits but if not used properly this can
be put to bad use quite easily. We are now going to talk about some pf the basic aspects
that are going to help us get a better idea of how scrapping is helpful but also throw light
on the flaws.

Data analysis is one of the key benefits that it helps us with. It can also help us with
visualisation of various aspects. Data analysis is a system and it has relevance in
every field or usable applications. It can be used in E-commerce, finance, IT or even
healthcare. This whole application also extends to its vital and important usage
everywhere. It has the possibility to help every business decision and provide
effectiveness to a lot of people in some way. Data analysis can’t be done without data
being provided to the system, therefore it needs to get data from processes like data
mining. It is the essential fuel that drives every analysis and data visualization

63 | P a g e
process. When it comes to data analysis, data from multiple sources is essential. This
kind of data especially requires high level of technical skills to collect, clean up and
organize. Web data scraping can be termed as an essential component of business
analysis now that more companies have grown their roots into the internet. The way
we used is one of its most common uses as it helps us to compare different features
between two products with respect to their price and everything. Industries always
want and need information that will help them know what they want to know of data.
Data helps in making really good and useful great business strategy no matter how
small your company is. Market analysis is how companies learn how to rise above
the requirement is while giving importance to the customers. Along with this, price
comparison can also be carried out using information that we can take from other
sites and know what needs to be selected on a better information of the products in
considerations. It also helps in RnD of products that the consumers may use once or
frequently or maybe an everyday use product. The market demands a better product
all the time as technology keeps improving whereas the utility decreases for the
product as it gets old therefore better stuff is demanded.. The development of better
products has to start from research. A lot of thought and exercises go into analysing
trends, demand and problems with the products that are already in the market and
need improvements to let the users feel that they are getting products every time they
buy something therefore this raises a need to developing them into better ones [41].
Research is an indispensable factor of product development and innovation. And, this
research needs huge amounts of data to be realised. Web data scraping has shown to
be one of the most helpful and useful procedure that gives the companies what they
require present day electronic gadgets.

Though data mining is not all good. It comes with a fair share of demerits that makes
one realise the important use of technology. Though all technologies that have been
developed need to be used ethically or in the wrong hands may [rove to be problem
causing for the society. when we talk about the demerits of data mining or scrapping
information directly from online links the first thing that comes to our mind is that it
may lead to use of information written by someone else in the form of Plagiarism[30].
Data scraping helps to take information for the internet in any form from all over the
World Wide Web in any form and keep it collected in one place. We agree that
information available online is free for use but this doesn’t mean that one can use it
for publishing on their own without permissions and other certifications required.
Doing so isn’t only something that would people frown upon because its unethical
but it is also illegal in most cases. One can file legal cases for his information being
used without permission and cause greater defamation towards the user. The next
demerit that could rise out of data mining is spamming. It may be one of the most
annoying things to come out of the internet. No one I say no one enjoys cold calls on
items we have liked or websites we have searched. There are companies that sell
products through these cold calls and they may keep bothering you with emails and
calls even after telling them five to six times that you aren’t interested in their
product. With the boom of the IT industry, spamming also rose as marketers realised
the potential of selling products over the phone or mail without the need to have a
massive physical store. One of the most common spammers are usually banks or
investment firms that want to sell you different schemes because they have deadlines

64 | P a g e
to meet and therefor keep pestering people with spam calls and messages. While we
are talking about all the problems that are caused one cannot deny that our personal
lives are at stake on the internet. We share so much data over the internet and if it
goes into the wrong hands one could easily use to steal our identity. We post so much
information over websites like Facebook and Instagram that one can actually very
easily retrieve all or contact details and further through spams and other unethical
issues choose to use it for various purposes. There one can agree that data mining
may be very helpful but can be very painful at times.

So after viewing so many merits and demerits of data mining we have seen that even
though it may raise so many problems for public it surely is very resourceful so in
the right hands it’s ethical and should not be frowned upon. Web data scraping is a
way in which the machine which is our computer goes to the internet and scraps the
information required on its own. We program this through different languages like
we personally tried it on python first and the on R-codes to see the effectiveness and
late use it for our project. to be honest there isn’t much difference between a person
using these codes or a machine automating the process the end result is gonna be the
same. But one should always read the terms and conditions of the website before
scrapping it because it may have certain warning or permission requirements. If you
follow the rules there is nothing that you need to worry about as you aren’t doing
anything wrong.

Ever since online retail has boomed one name that comes up on the top is amazon.
Amazon is now the leading company worldwide that sells products online, its revenue
has grown in multi-level folds over the years. It’s easy to use and very fruitful for the
seller and purchasers both. One can post all sorts of reviews over the website so we
scrapped these reviews to get more information about the products that were in
consideration and use them for better understanding of what the customers feel about
our product.
Even though there are various online data scrapping tool available on the internet that
just require you to submit the link. We extracted comments from amazon and these
comments are in personal form over the website, this is depicted in figure 15. We took
a conventional approach towards this and used python and R-codes for our study. R is
a programming language and free software environment for statistical computing e

have disadvantage instances, delineate as vectors of feature

values, where the class labels unit per line or per process drawn from some finite
set. There is not one algorithm for work, they are unique such classifiers, but a family

65 | P a g e
of algorithms supported or gained a typical principle: all naive Bayes classifiers
assume that the value or the result or estimation of a particular feature is freelance
of the value of the opposite feature, given the class variable. As associate
degree example, a fruit might even be thought of to be associate degree apple if it's
and graphics supported by the R Foundation for Statistical Computing. The R language is
widely used among statisticians and data miners for developing statistic al software and data
analysis.

Fig 8- the source of source website used is amazon

Amazon knows how the industry works so it self provides APIs that help people scrap data
from the source. Figure 8 shows the screen we are interested in which contains the reviews.
But now these don’t get us all the required data but it everything that the amazon website is
giving permission to be scrapped from the website as per its wishes. These are product
advertising APIs that make this whole procedure very convenient. Using these we can extract
various aspect of information from the website. We personally we interested in ratings and
reviews of products. So to start scrapping we first need to get the ASIN codes for the products
which is retrieved from amazon itself. A customer’s reviews towards a product or service open
a vast horizon for improvement of the good or service as it gives a major insight about the
personal thoughts of customer. So the industry aims to make products that suit their customers
better as this wo uld in turn help them increase their market share as their product will end up
being better to other products available in the market. Once a product achieves a core
competence for being better it the revenue goes up as the sales increase. This proves that how
critical feedback data of customer is to an industry. Sentiment analysis can be done on a vast
variety of application. But in this regard we talk about it being used to scrap and analyse
reviews of products. Particularly talking about reviews about ph ones and tablets. With the
boom of market now we have a wide variety of phones and tablets available in stores. They
vary in configurations and features and each product delivers some or the other aspect with

66 | P a g e
arises different sorts of emotions in customers . Reviews consist of these emotions that help
the companies understand their products market stand. This was done for the different phones
in consideration. We build the URL, and point to a specific node

#productTitle of the HTML web page using the CSS selector

67 | P a g e
Naive Bayes technique for constructing classifications and categories that are required
for some or the other purpose models that assign category that let them have labels
red, round, and regarding 10 cm in diameter. A naïve Bayes classifier
considers each of these choices or options to contribute severally to the chance that
this fruit or result od the provided options is associate degree apple, despite
any possible correlations between the colour, roundness, and diameter choices but
because we are talking about phones so we see the features. For some sorts
of chance models, and n most cases naïve Bayes classifiers is
trained really expeditiously throughout a well-known and decently
supervised learning setting. In many smart applications, parameter estimation for
naive man of science models uses the manoeuvre of most chance. In
numerous words, one can work with the naive Bayes model whereas not allowance
and the possibilities of all the rejection theorem chance or practice any or process
theorem ways. Despite their naive Bayes and apparently easy assumptions,
naive Bayes classifiers have worked quite well in all the advanced realworld things.
An advantage of naïve Bayes is that it exclusively wants a tiny
low vary of work info to estimate the parameters necessary for
classification.

68 | P a g e
website.

First step was data collection where we scraped data from Amazon It felt
contained reviews from customers on iPhone 10 and one plus 6 product
delivery. These contained all the positive negative or neutral that the like
customers discussing through the reviews. Next we used R code software
to do text mining where frequently appeared words were tabulated. These
words were things that the people talked about over the reviews again and
again. This gives an insight on what the people were expecting and what
they received. The tabulated words were used to plot word cloud to get a
pictorial representation of the result. Word clouds give explanation of the a visual
rev iews and what can be done or what’s being done well. Analysing the
word cloud, we could analyse the better product delivery among the two
flagship companies.

Through a brief look at the code one can understand its processings. When
and how a library were opened and introduced. This also shows how the
documents that were developed from the scrapping codes are now
introduced and used.

As previously discussed and shown that our results a sets of word clouds of the
for the products that have been retrieved from the mining and sentiment
analysis products review. Now talking about the benefits and pro of the
formed word clouds is that they help us get a visual representation of the
data that is written again and again in the reviews by customers. This gets
an in sight into the mindset of the customers. customer’s reviews towards
a product or service open a vast horizon for improvement of the good or
service as it gives a major insight about the personal thoughts of customer.
So the industry aims to make product s that suit their customers better as
this would in turn help them increase their market share as their product
will end up being A
better to other products available in the market. Once a product achieves a core
competence for being better it the revenue goes up as the sales increase. This proves
that how critical feedback data of customer is to an industry. Sentiment analysis can
be done on a vast variety of application. But in this regard we talk about it being used
to scrap and analyse reviews of products. Particularly talking about reviews about
phones and tablets. With the boom of market now we have a wide variety of phones
and tablets available in stores. They vary in configurations and features and each
product delivers some or the other aspect with arises different sorts of emotions in
customers. Reviews consist of these emotions that help the companies understand their
products market stand.

69 | P a g e
Now that we have the word clouds we are going to analyse what the customer’s
emotions about the product sin consideration tell about the what should be done to
improve the product. Every industry works with a moto of increasing profits as it at
the end of the day that is what pays for the bread and butter of everyone involved in
the industry. This is achieved by when the customers of the company, the ones who
are paying and helping the company generate revenue are satisfied with the goods and
services provided to them. A satisfied customer tends to spend more and show loyalty
towards the company. The following figures figure 6 and figure 6 are going to depict
the word clouds retrieved from the analysis of Oneplus reviews and iPhone reviews.
This satisfaction of customers is very critical to the company. Therefore analysing
these word clouds is very important and knowing what is to be done by them later on
is derived on the analysis of this.

Fig 9 word frequency generated

So here we tend to see the words that are recurrent the utmost range of times. This
has been done by the system. Figure 9 shows the word frequencies developed by
the
system. The system has recognised the amount of times a word has been used
and depicted it within the manner of an inventory. we tend to see that for the
One and phone word quality has been used many times and additionally verified.
this
implies that the purchasers appear happy by the merchandise as they appear to
love the standard of the phone and area unit happy. The attributes like smart and
verified show that the purchasers area unit happy which the merchandise is sweet.
is

70 | P a g e
This additionally tells the manufacturer a feedback that his product quality t
sweet and he has to continue maintaining these standards for
gaining client loyalty. individuals talked regarding the charging and battery life for
this phone. This shows that folks area unit happy with the battery life. If
One and introduced higher and larger batteries in its phones that may be
charged terribly quickly. Their newer technologies regarding dash charging appears
have caught on interest with the purchasers. This feedback tell s the manufacturer that
his innovation has helped him gain competitive advantage over the
opposite brands. this can be important to achieve market
leadership.
one and continues the great work it's going to have a awfully bright future. we tend
to see that for the One and phone word purchase has been used many times
and additionally verified. this implies that the purchasers appear happy by the
merchandise as they appear to love the standard of the phone and area unit happy. The
attributes like nice and worth show that the purchasers area unit happy which the
merchandise is sweet. This additionally tells the manufacturer a feedback that
If
his product quality is sweet and he has to continue maintaining these standards for
gaining client loyalty. individuals talked regarding the charging and battery life for
this phone. This shows that folks area unit happy with the battery life.
One and introduced higher and larger cameras in its phones that have terribly high
resolution. Their newer technologies regarding appear cameras appears to
own caught on interest with the purchasers. This feedback tells the manufacturer that
his innovation has helped him gain competitive advantage over the
opposite brands. this can be important to achieve market
leadership.
one and continues the great work it's going to have a awfully bright future.

On clients hand may be a customer sees that one and phone’s word frequency he or
she might realise that what the other customers feel regarding the merchandise and he
himself are able to build a wiser call regarding the phone. We tend to see that for the
One and phone word quality has been used many times and additionally verified. this
suggests that the opposite users appear happy by the merchandise as they appear to
love the standard of the phone and ar glad. The attributes like smart and verified show
that the opposite users ar glad which the merchandise is nice. This additionally tells
the new client that merchandise quality is nice and selecting this one might show
fruitful future. Users talked regarding the charging and battery life for this phone. This
shows that folks ar glad with the battery life. One and introduced higher and
larger batteries in its phones that may be charged terribly quickly. Their newer
technologies regarding dash charging looks to possess caught on interest with all
users. This data tells the new client that the merchandise has advantage over the
opposite brands. we tend to see that for the One and phone word purchase has been
used many times and additionally verified. this suggests that the

71 | P a g e
purchasers appear happy by the merchandise as they appear to love the standard of
the phone and ar glad. The attributes like nice and worth show that the
purchasers ar glad which the merchandise is nice.. individuals talked regarding the
charging and battery life for this phone. This shows that folks ar glad with the battery
life. One and introduced higher and larger cameras in its phones that
have terribly high resolution. Their newer
technologies regarding pop cameras looks to possess caught on interest with the
purchasers. This tells the client that the merchandise if technically forward and
selecting it's going to be a god call.

Now when we talk about the iPhone when we see that word frequencies for the iPhone
we see find various kinds of information available to us. We see that for the iPhone
word expensive has been used several times and also verified. This means that the
customers seem worried that the product is very costly and may not be value for money.
But they are satisfied with the quality as the iPhone is a well optimised product. The
attributes like good and dependable show that the customers are satisfied and that the
product is good. This also tells the manufacturer a feedback that his products quality is
good and he needs to continue maintaining these standards for gaining customer
loyalty. People talked about the display and screen size for this phone. This shows that
people are satisfied with the screen. IPhone introduced higher and bigger screens in its
phones that have high resolutions. Their newer technologies about OLED seems to
have caught on interest with the customers. This feedback tells the manufacturer that
his innovation has helped him gain competitive advantage over the other brands. This
is very important to gain market leadership. If one plus continues the good work it may
have a very bright future. We see that for the iPhone word color has been used several
times. This means that the customers seem pleased by the product as they seem to like
the quality of the phone and are satisfied. The attributes like grey and silver show that
the customers are satisfied and like these colours on the phone. This also tells the
manufacturer a feedback that his products quality is good and he needs to continue
maintaining these standards for gaining customer loyalty. People talked about the
music system for this phone. This shows that people are satisfied with the music
system. IPhone introduced airpods with their new phone and removed the AUX jack.
Their newer technologies about airpods seems to have caught on interest with the
customers. This feedback tells the manufacturer that his innovation has helped him
gain competitive advantage over the other brands. This is very important to gain market
leadership. If one plus continues the good work it may have a very bright future.

72 | P a g e
users seem worried that the product is very costly and may not be value for money. But
they are satisfied with the quality as the iPhone is a well optimised product. The
attributes like good and dependable show that the users are satisfied and that the
product is good. This also tells the new customer that the product is dependable and
buying it seems like a good idea. People talked about the display and screen size for
this phone. This shows that people are satisfied with the screen. IPhone introduced
higher and bigger screens in its phones that have high resolutions. Their newer
technologies about OLED seems to have caught on interest with all the users. This
information tells the customer that the product seems good as compared to the other
products. This is very important to gain market leadership. If apple continues the good
work it may have a very bright future. We see that for the iPhone word colour has been
used several times. This means that the customers seem pleased by the product as they
seem to like the quality of the phone and are satisfied. The attributes like grey and
silver show that the customers are satisfied and like these colours on the phone. This
tells the new user that most people opt for Grey and Silver colours. So if he wants his
product to look like the others he can also choose one of these colours. But if he wants
a certain degree of uniqueness in the product he can opt for other colours available.
People talked about the music system for this phone. This shows that people are
satisfied with the music system. IPhone introduced airpods with their new phone and
removed the AUX jack. Their newer technologies about airpods seems to have caught
on interest with the customers. This information tells the new customer that buying this
product seems like a good idea as they keep experimenting with new technologies.
This way the system helps both the manufacturers and even the users to make better
decisions when it comes to these phones.

Fig: 10- iPhone world cloud from its reviews

So clearly from the word map we can see what the customers emotions tell us about
their requirement and expectations being met by the product. In figure 10 We see that
most people comment about the colours of the product. The phone comes has three
variants namely space grey, silver and golden. People have mostly preferred the space
grey variant therefore we see that this is a preferred choice so the manufacture should

73 | P a g e
not discontinue this color option in later models. Similarly people have talked about
the screen. The screen has always been believed to be better on iPhone but these
complaints about the camera in the phone tells the manufactures to make some
improvements. One of the major concerns that is shown by the word cloud is that the
product is expensive. A person’s utility is also variable to the cost he has paid for the
product. So a costly product should drive a lot higher utility but this doesn’t seem to
be the case for the phone. Therefore the company should look for ways to reduce the
costs of its products for its users. Comments can also be seen about the design and
weight of the product. The IPhone is very tastefully made and is very light weight. This
means that the customers are happy with the fact that the product is providing these
promises. People even talk about the optimization of the product. The phone doesn’t
usually lag and promises a very long life without much troubles over the years. .
Similarly people have talked about the camera. The camera has always been believed
to be better on iPhone but these complaints about the camera in the phone tells the
manufactures to make some improvements. One of the major concerns that is shown
by the word cloud is that the product is expensive. A person’s utility is also variable to
the cost he has paid for the product. So a costly product should drive a lot higher utility
but this doesn’t seem to be the case for the phone. The display has always been believed
to be better on iPhone but these complaints about the camera in the phone tells the
manufactures to make some improvements. The speaker has always been believed to
be better on iPhone but these complaints about the camera in the phone tells the
manufactures to make some improvements. These inferences help the company know
what the customers feel about the product in one glance and therefore they can work
on the short comings and improve the product for a higher customer satisfaction.

Fig: 11- word cloud from One plus

When we look at the word cloud for one plus phone. Figure 11 has been derived by the
system as the word cloud for One plus. There are many things that can be inferred.
People have raised concerns about the screen size being too big. It goes beyond the
comfort of the palm of a person. The company should work o improving the

74 | P a g e
ergonomics of the phone. People compliment the camera of the phone as it is seen in
so many pictures. The phone is called value for money as it is a fraction of the cost of
the other phone in consideration. As the costs are low the customer satisfactions are
high. People also talk about the face unlock and how it cases errors in the dark places
this can be very painful for the user and a game changer when it comes to a person’s
loyalty to the company. A company can’t afford to lose its loyal customers as it reduces
revenues and the general market opinion of the company. So the company should see
to it to correct these errors so that the customer remain satisfied. People also talk about
the speaker and how it cases errors in the crowded places this can be very painful for
the user and a game changer when it comes to a person’s loyalty to the company. A
company can’t afford to lose its loyal customers as it reduces revenues and the general
market opinion of the company[18]. So the company should see to it to correct these
errors so that the customer remain satisfied. People have raised concerns about the
display being dark for use in sunlight. It goes beyond the comfort of human eye. The
company should work o improving the ergonomics of the phone. People compliment
the charging speed of the phone as it is seen in so many pictures. The company provides
different innovation for the speed of charging as they saw that people require this. This
now has become one of the core competency for the company. Similarly people have
talked about the camera. The bluetooth has always been believed to be better on
Oneplus but these complaints about the Bluetooth technology in the phone tells the
manufactures to make some improvements. One of the major concerns that is shown
by the word cloud is that the product is expensive. A person’s utility is also variable to
the cost he has paid for the product. These inferences help the company know what the
customers feel about the product in one glance and therefore they can work on the short
comings and improve the product for a higher customer satisfaction.

75 | P a g e
6-CONCLUSION AND SCOPE FOR BETTER WORK

The main objective of this project was to create a platform that is unbiased of the
predicaments of the buyers who pay websites to show good ratings of their products.
This type of analysis opens the scope for, call recording and analysing the recorded
conversation and predicting about the customer’s interest in a particular product. After
facing the same problem day in and day out over where to buy the product from, it is
needed to make a standardize effort on identifying what can be the determining
parameter to address the problem. This project intends to deal with the fore-mentioned
problem. The customer reviews on websites of amazon, flipkart, snap deal and many
more, post their thoughts on the product delivery, it is hard to go through all the text
and reviews to identify the results one is looking for. Sentiment analysis gives a list of

76 | P a g e
word frequencies that is repeated in a set of text, which gives a basic outlook on which
a ground level conclusion can be made.

We learnt various approaches of how to identify the key critical remarks over a
sentiment was validated in the papers[50]. The scope of extending the outlooks from
specific reviews to twitter tweets can also be used as a way to collect the data [15][27]. In
one of the papers, there key way to analyze sentiments were to use automated opinion
mining and use it to detect sarcasm. First possible step in solving any problem related
to the issue is to generalize pieces of common senses knowledge in terms of conceptual
primitives that could catch most semantic inflections of natural language concepts [51].
Determining the polarity of a text, it is also difficult to extract the negative or positive
utterances on a more fined grained level, so it’s important to identify such words while
doing word mining. Various probability theorems were taken in consideration to
predict the nature of sentiments. Use of weighted scheme from information retrieval to
enhance accuracy of sentimental analysis.

There exist some possible extensions to our work. It would be interesting if we can
compare different products based on various parameters and help the user decide with
one is better. For Example, we could compare IPhone and OnePlus mobile phones
based of the technology offered, aesthetics, Value/Price etc. and help the customer
understand with product is a better deal and from where they should buy it based to the
service offered by the company which we determine using sentiment analysis from
customer reviews[25][10]. In addition to that we look forward to use other platform like
twitter, Facebook etc[52]. to gather view about the company’s services and also perform
a sentiment classification on hashtags as we believe this is significant for sentiment
analysis of topics since hashtags can be nearly viewed as user-annotated topics.

77 | P a g e
7-REFERENCES

[1]Kouloumpis, E., Wilson, T. and Moore, J., 2011, July. Twitter sentiment analysis:
The good the bad and the omg!. In Fifth International AAAI conference on weblogs
and social media.

[2]Wang, X., Wei, F., Liu, X., Zhou, M. and Zhang, M., 2011, October. Topic
sentiment analysis in twitter: a graph-based hashtag sentiment classification approach.
In Proceedings of the 20th ACM international conference on Information and
knowledge management (pp. 1031- 1040). ACM.

[3]Ravi, K. and Ravi, V., 2015. A survey on opinion mining and sentiment analysis: tasks,
approaches and applications. Knowledge-Based Systems, 89, pp.14-46.

[4]Paltoglou, G. and Thelwall, M., 2010, July. A study of information retrieval

weighting schemes for sentiment analysis. In Proceedings of the 48th annual meeting
of the association for computational linguistics (pp. 1386-1395). Association for
Computational Linguistics.

[5] Wang, H., Can, D., Kazemzadeh, A., Bar, F. and Narayanan, S., 2012, July. A
system for real-time twitter sentiment analysis of 2012 us presidential election
cycle. In Proceedings of the ACL 2012 system demonstrations (pp. 115-120).
Association for Computational Linguistics.

78 | P a g e
[6] Mullen, T. and Collier, N., 2004. Sentiment analysis using support vector
machines with diverse information sources. In Proceedings of the 2004 conference
on empirical methods in natural language processing (pp. 412-418).

[7] Kasper, W. and Vela, M., 2011, October. Sentiment analysis for hotel reviews. In
Computational linguistics-applications conference (Vol. 231527, pp. 45-52).

[8] Guzman, E., Azócar, D. and Li, Y., 2014, May. Sentiment analysis of commit
comments in GitHub: an empirical study. In Proceedings of the 11th Working
Conference on Mining Software Repositories (pp. 352-355). ACM.
[9] Bae, Y. and Lee, H., 2012. Sentiment analysis of twitter audiences: Measuring the
positive or negative influence of popular twitterers. Journal of the American
Society for Information Science and Technology, 63(12), pp.2521-2535.

[10] Basile, V., 2013, June. Sentiment analysis on Italian tweets. In Proceedings of the
4th Workshop on Computational Approaches to Subjectivity, Sentiment and
Social Media Analysis (pp. 100-107).

[11] Duwairi, R.M., Marji, R., Sha'ban, N. and Rushaidat, S., 2014, April. Sentiment
analysis in arabic tweets. In 2014 5th International Conference on Information and
Communication Systems (ICICS) (pp. 1-6). IEEE.

[12] Hao, M., Rohrdantz, C., Janetzko, H., Dayal, U., Keim, D.A., Haug, L.E. and Hsu,
M.C., 2011, October. Visual sentiment analysis on twitter data streams. In 2011

IEEE Conference on Visual Analytics Science and Technology (VAST) (pp. 277278).
IEEE.

[13] Liu, B. and Zhang, L., 2012. A survey of opinion mining and sentiment analysis.
In Mining text data (pp. 415-463). Springer, Boston, MA.

[14] Cambria, E., Poria, S., Bajpai, R. and Schuller, B., 2016, December. SenticNet 4:
A semantic resource for sentiment analysis based on conceptual primitives. In
Proceedings of COLING 2016, the 26th international conference on
computational linguistics: Technical papers (pp. 2666-2677).

79 | P a g e
[15] Maynard, D.G. and Greenwood, M.A., 2014, March. Who cares about sarcastic
tweets? investigating the impact of sarcasm on sentiment analysis. In LREC 2014
Proceedings. ELRA.

[16] Pang, B. and Lee, L., 2008. Opinion mining and sentiment analysis. Foundations
and Trends® in Information Retrieval, 2(1–2), pp.1-135.

[17] Wilson, T., Wiebe, J. and Hoffmann, P., 2005. Recognizing contextual polarity in
phraselevel sentiment analysis. In Proceedings of Human Language Technology
Conference and Conference on Empirical Methods in Natural Language
Processing.

[18] Pak, A. and Paroubek, P., 2010, May. Twitter as a corpus for sentiment analysis
and opinion mining. In LREc (Vol. 10, No. 2010, pp. 1320-1326).
[19] Pang, B. and Lee, L., 2004, July. A sentimental education: Sentiment analysis
using subjectivity summarization based on minimum cuts. In Proceedings of the
42nd annual meeting on Association for Computational Linguistics (p. 271).
Association for Computational Linguistics.

[20] Liu, B., 2010. Sentiment analysis and subjectivity. Handbook of natural language
processing, 2(2010), pp.627-666.

[21] Kouloumpis, E., Wilson, T. and Moore, J., 2011, July. Twitter sentiment analysis:
The good the bad and the omg!. In Fifth International AAAI conference on
weblogs and social media.

[22] Wang, X., Wei, F., Liu, X., Zhou, M. and Zhang, M., 2011, October. Topic
sentiment analysis in twitter: a graph-based hashtag sentiment classification
approach. In Proceedings of the 20th ACM international conference on
Information and knowledge management (pp. 1031- 1040). ACM.

[23] Ravi, K. and Ravi, V., 2015. A survey on opinion mining and sentiment analysis:
tasks, approaches and applications. Knowledge-Based Systems, 89, pp.14-46.

80 | P a g e
[24] Paltoglou, G. and Thelwall, M., 2010, July. A study of information retrieval
weighting schemes for sentiment analysis. In Proceedings of the 48th annual
meeting of the association for computational linguistics (pp. 1386-1395).
Association for Computational Linguistics.

[25] Wang, H., Can, D., Kazemzadeh, A., Bar, F. and Narayanan, S., 2012, July. A
system for real-time twitter sentiment analysis of 2012 us presidential election
cycle. In Proceedings of the ACL 2012 system demonstrations (pp. 115-120).
Association for Computational Linguistics.

[26] Mullen, T. and Collier, N., 2004. Sentiment analysis using support vector
machines with diverse information sources. In Proceedings of the 2004 conference
on empirical methods in natural language processing (pp. 412-418).

[27] Go, A., Bhayani, R. and Huang, L., 2009. Twitter sentiment classification using
distant supervision. CS224N Project Report, Stanford, 1(12), p.2009.
[28] Kaji, N. and Kitsuregawa, M., 2007, June. Building lexicon for sentiment analysis
from massive collection of HTML documents. In Proceedings of the 2007 Joint
Conference on Empirical Methods in Natural Language Processing and
Computational Natural Language Learning (EMNLP-CoNLL) (pp. 1075-1083).

[29] Abdul-Mageed, M., Diab, M.T. and Korayem, M., 2011, June. Subjectivity and
sentiment analysis of modern standard Arabic. In Proceedings of the 49th Annual
Meeting of the Association for Computational Linguistics: Human Language
Technologies: short papersVolume 2 (pp. 587-591). Association for
Computational Linguistics.

[30] Mittal, A. and Goel, A., 2012. Stock prediction using twitter sentiment analysis.
Standford University, CS229 (2011 http://cs229. stanford.
edu/proj2011/GoelMittalStockMarketPredictionUsingTwitterSentimentAnalysis.
pdf), 15.

81 | P a g e
[31] Brooke, J., Tofiloski, M. and Taboada, M., 2009, September. Cross-linguistic
sentiment analysis: From English to Spanish. In Proceedings of the international
conference RANLP2009 (pp. 50-54).

[32] Neethu, M.S. and Rajasree, R., 2013, July. Sentiment analysis in twitter using
machine learning techniques. In 2013 Fourth International Conference on
Computing, Communications and Networking Technologies (ICCCNT) (pp. 1-5).
IEEE.

[33] Wöllmer, M., Weninger, F., Knaup, T., Schuller, B., Sun, C., Sagae, K. and
Morency, L.P., 2013. Youtube movie reviews: Sentiment analysis in an audio-
visual context. IEEE Intelligent Systems, 28(3), pp.46-53.

[34] Zirn, C., Niepert, M., Stuckenschmidt, H. and Strube, M., 2011, November.
Finegrained sentiment analysis with structural features. In Proceedings of 5th
International Joint Conference on Natural Language Processing (pp. 336-344).

[35] Poria, S., Cambria, E., Winterstein, G. and Huang, G.B., 2014. Sentic patterns:
Dependency-based rules for concept-level sentiment analysis. Knowledge-Based
Systems, 69, pp.45-63.
[36] Yu, Y., Duan, W. and Cao, Q., 2013. The impact of social and conventional media
on firm equity value: A sentiment analysis approach. Decision Support Systems,
55(4), pp.919-926.

[37] Mullen, T. and Malouf, R., 2006, March. A Preliminary Investigation into
Sentiment Analysis of Informal Political Discourse. In AAAI Spring Symposium:
Computational Approaches to Analyzing Weblogs (pp. 159-162).

[38] Tan, S., Cheng, X., Wang, Y. and Xu, H., 2009, April. Adapting naive bayes to
domain adaptation for sentiment analysis. In European Conference on Information
Retrieval (pp. 337- 349). Springer, Berlin, Heidelberg.

82 | P a g e
[39] Guzman, E. and Maalej, W., 2014, August. How do users like this feature? a fine
grained sentiment analysis of app reviews. In 2014 IEEE 22nd international
requirements engineering conference (RE) (pp. 153-162). IEEE.

[40] Zhang, Y., Lai, G., Zhang, M., Zhang, Y., Liu, Y. and Ma, S., 2014, July. Explicit
factor models for explainable recommendation based on phrase-level sentiment
analysis. In Proceedings of the 37th international ACM SIGIR conference on
Research & development in information retrieval (pp. 83-92). ACM.

[41] Bautin, M., Vijayarenu, L. and Skiena, S., 2008, April. International sentiment
analysis for news and blogs. In ICWSM.

[42] Boiy, E., Hens, P., Deschacht, K. and Moens, M.F., 2007, June. Automatic
Sentiment Analysis in On-line Text. In ELPUB (pp. 349-360).

[43] Abdul-Mageed, M., Diab, M.T. and Korayem, M., 2011, June. Subjectivity and
sentiment analysis of modern standard Arabic. In Proceedings of the 49th Annual
Meeting of the Association for Computational Linguistics: Human Language
Technologies: short papersVolume 2 (pp. 587-591). Association for
Computational Linguistics.

[44] Go, A., Bhayani, R. and Huang, L., 2009. Twitter sentiment classification using
distant supervision. CS224N Project Report, Stanford, 1(12), p.2009.

[45] Li, F., Huang, M. and Zhu, X., 2010, July. Sentiment analysis with global topics
and local dependency. In Twenty-Fourth AAAI Conference on Artificial
Intelligence. [46] Zhang, Y., Lai, G., Zhang, M., Zhang, Y., Liu, Y. and Ma, S.,
2014, July. Explicit factor models for explainable recommendation based on
phrase-level sentiment analysis. In Proceedings of the 37th international ACM
SIGIR conference on Research & development in information retrieval (pp. 83-
92). ACM.

[47] Pang, B. and Lee, L., 2008. Opinion mining and sentiment analysis. Foundations
and Trends® in Information Retrieval, 2(1–2), pp.1-135.

83 | P a g e
[48] Wilson, T., Wiebe, J. and Hoffmann, P., 2005. Recognizing contextual polarity in
phraselevel sentiment analysis. In Proceedings of Human Language Technology
Conference and Conference on Empirical Methods in Natural Language Processing.

[49] Pak, A. and Paroubek, P., 2010, May. Twitter as a corpus for sentiment analysis
and opinion mining. In LREc (Vol. 10, No. 2010, pp. 1320-1326).

[50] Pang, B. and Lee, L., 2004, July. A sentimental education: Sentiment analysis
using subjectivity summarization based on minimum cuts. In Proceedings of the 42nd
annual meeting on Association for Computational Linguistics (p. 271). Association for
Computational Linguistics.

[51] Liu, B., 2010. Sentiment analysis and subjectivity. Handbook of natural language
processing, 2(2010), pp.627-666.

[52] Kouloumpis, E., Wilson, T. and Moore, J., 2011, July. Twitter sentiment analysis:
The good the bad and the omg!. In Fifth International AAAI conference on weblogs
and social media.

84 | P a g e

The Knowledge Graph Cookbook
No ratings yet
The Knowledge Graph Cookbook
228 pages
Scan To BIM - Presentation
No ratings yet
Scan To BIM - Presentation
61 pages
The Hundred Page Machine Learning Book
No ratings yet
The Hundred Page Machine Learning Book
7 pages
The Power of Data Storytelling by Sejal Vora 2019 9789353282905 9789353282912 Compress
No ratings yet
The Power of Data Storytelling by Sejal Vora 2019 9789353282905 9789353282912 Compress
249 pages
Data Visualization With Pandas
No ratings yet
Data Visualization With Pandas
8 pages
Data Science Project Report
100% (1)
Data Science Project Report
3 pages
Building Recommendation System Using Movielens Data
No ratings yet
Building Recommendation System Using Movielens Data
6 pages
637122053082671964عبدالله سالم محمد الراشدي PDF
0% (1)
637122053082671964عبدالله سالم محمد الراشدي PDF
2 pages
Twitter Sentiment Analysis
No ratings yet
Twitter Sentiment Analysis
5 pages
Sentiment Analysis Over Online Product Reviews A Survey
No ratings yet
Sentiment Analysis Over Online Product Reviews A Survey
9 pages
Data Science in E-Commerce - Report - Writing
No ratings yet
Data Science in E-Commerce - Report - Writing
18 pages
Twitter Sentiment Analysis
No ratings yet
Twitter Sentiment Analysis
4 pages
Database Management Systems by Raghu Ramakrishnan: Special Features of Book
No ratings yet
Database Management Systems by Raghu Ramakrishnan: Special Features of Book
3 pages
Sentimental Analysis Final Year Project
No ratings yet
Sentimental Analysis Final Year Project
21 pages
20 Machine Learning Projects For Beginners
No ratings yet
20 Machine Learning Projects For Beginners
22 pages
Introduction To Data Science
No ratings yet
Introduction To Data Science
200 pages
Sentiment Analysis
No ratings yet
Sentiment Analysis
5 pages
Role of Machine Learning in The Field of Fiber Reinforced Polymer
No ratings yet
Role of Machine Learning in The Field of Fiber Reinforced Polymer
6 pages
Ensemble Learning: Wisdom of The Crowd
100% (1)
Ensemble Learning: Wisdom of The Crowd
12 pages
MCA Project Titles
No ratings yet
MCA Project Titles
2 pages
Big Data Analysis
100% (1)
Big Data Analysis
30 pages
Computational Intelligent Data Analysis For Sustainable Development PDF
No ratings yet
Computational Intelligent Data Analysis For Sustainable Development PDF
443 pages
M.tech - Artificial Intelligence and Data Science
No ratings yet
M.tech - Artificial Intelligence and Data Science
57 pages
Augmented Analytics
No ratings yet
Augmented Analytics
8 pages
Lecture 01 (Introduction To Pattern Recognition)
No ratings yet
Lecture 01 (Introduction To Pattern Recognition)
26 pages
"Sentiment Analysis of Imdb Movie Reviews": A Project Report
No ratings yet
"Sentiment Analysis of Imdb Movie Reviews": A Project Report
27 pages
Introduction To Data Mining
100% (1)
Introduction To Data Mining
643 pages
Big Data Analytics in Weather Forecasting
No ratings yet
Big Data Analytics in Weather Forecasting
29 pages
2018 Book DataScienceAndPredictiveAnalyt
No ratings yet
2018 Book DataScienceAndPredictiveAnalyt
929 pages
Mall Customer Data Analysis PDF
No ratings yet
Mall Customer Data Analysis PDF
10 pages
Sentiment Analysis On Movie Reviews: Natural Language Processing UML602 Project Report
No ratings yet
Sentiment Analysis On Movie Reviews: Natural Language Processing UML602 Project Report
13 pages
Semantic Web Technologies For Intelligent Engineering Applications
No ratings yet
Semantic Web Technologies For Intelligent Engineering Applications
413 pages
Red Wine Quality Detection
No ratings yet
Red Wine Quality Detection
17 pages
Sentimental Analysis of Movie Review
100% (1)
Sentimental Analysis of Movie Review
58 pages
5. CHAPTER 2
No ratings yet
5. CHAPTER 2
7 pages
Deep Learning Project
No ratings yet
Deep Learning Project
24 pages
BA ZG523 Introduction To Data Science
100% (1)
BA ZG523 Introduction To Data Science
12 pages
Anomaly Detection in Images CIFAR-10
No ratings yet
Anomaly Detection in Images CIFAR-10
9 pages
An Application of Selected Artificial Intelligence Techniques To Engineering Analysis
No ratings yet
An Application of Selected Artificial Intelligence Techniques To Engineering Analysis
153 pages
Resume - Rajat Chaturvedi
No ratings yet
Resume - Rajat Chaturvedi
3 pages
Sentiment Analysis
No ratings yet
Sentiment Analysis
5 pages
Big Data and Social Media Analytics Trending Applications by Mehmet Çakırtaş, Mehmet Kemal Ozdemir
No ratings yet
Big Data and Social Media Analytics Trending Applications by Mehmet Çakırtaş, Mehmet Kemal Ozdemir
246 pages
A Survey On Opinion Mining and Sentiment Analysis: Tasks, Approaches and Applications1-S2.0-S0950705115002336-Main
No ratings yet
A Survey On Opinion Mining and Sentiment Analysis: Tasks, Approaches and Applications1-S2.0-S0950705115002336-Main
33 pages
ProdEconR PDF
No ratings yet
ProdEconR PDF
370 pages
Data Analytics
No ratings yet
Data Analytics
134 pages
Technical Analysis Course Syllabus
No ratings yet
Technical Analysis Course Syllabus
1 page
Twitter Sentiment Analysis With Textblob
No ratings yet
Twitter Sentiment Analysis With Textblob
6 pages
Complete Statistical Methods For Machine Learning: Discover How To Transform Data Into Knowledge With Python Jason Brownlee PDF For All Chapters
100% (2)
Complete Statistical Methods For Machine Learning: Discover How To Transform Data Into Knowledge With Python Jason Brownlee PDF For All Chapters
62 pages
A Hybrid CNN-LSTM: A Deep Learning Approach For Consumer Sentiment Analysis Using Qualitative User-Generated Contents
No ratings yet
A Hybrid CNN-LSTM: A Deep Learning Approach For Consumer Sentiment Analysis Using Qualitative User-Generated Contents
15 pages
Advances in Machine Learning and Signal Processing
100% (1)
Advances in Machine Learning and Signal Processing
309 pages
103 RO4 Final 201819
No ratings yet
103 RO4 Final 201819
124 pages
Sentiments Analysis Using Ai: Project Report
No ratings yet
Sentiments Analysis Using Ai: Project Report
27 pages
Lab 3 - Linear Regression
No ratings yet
Lab 3 - Linear Regression
15 pages
Big Data: Opportunities, Strategy and Challenges
100% (1)
Big Data: Opportunities, Strategy and Challenges
17 pages
Towards Geometric Deep Learning I - On The Shoulders of Giants
No ratings yet
Towards Geometric Deep Learning I - On The Shoulders of Giants
13 pages
DataMining Lecture 1
No ratings yet
DataMining Lecture 1
35 pages
Paper 1-Bidirectional LSTM With Attention Mechanism and Convolutional Layer
100% (1)
Paper 1-Bidirectional LSTM With Attention Mechanism and Convolutional Layer
51 pages
Sentiment Analysis of Product Review
No ratings yet
Sentiment Analysis of Product Review
6 pages
DataScience Interview Questions
100% (1)
DataScience Interview Questions
66 pages
Twitter Sentiment Analysis For Product Reviews To Gather Information Using Machine Learning Technique
No ratings yet
Twitter Sentiment Analysis For Product Reviews To Gather Information Using Machine Learning Technique
6 pages
How to Research Qualitatively: Tips for Scientific Working
From Everand
How to Research Qualitatively: Tips for Scientific Working
Martin Gertler
No ratings yet
7th Sem Report File
No ratings yet
7th Sem Report File
41 pages
BDD Vista 2007
No ratings yet
BDD Vista 2007
56 pages
Annotated Bibliography of Animation
No ratings yet
Annotated Bibliography of Animation
3 pages
Balazsi-Matyas: The Estimation of Multi-Dimensional FixedEffects Panel Data Models
No ratings yet
Balazsi-Matyas: The Estimation of Multi-Dimensional FixedEffects Panel Data Models
38 pages
901 Cpu Book Cp3 Cpu en
No ratings yet
901 Cpu Book Cp3 Cpu en
167 pages
History of Excel
No ratings yet
History of Excel
1 page
Webinar
No ratings yet
Webinar
16 pages
[Ebooks PDF] download Growing Algorithms and Data Structures 4th Edition David Scuse full chapters
No ratings yet
[Ebooks PDF] download Growing Algorithms and Data Structures 4th Edition David Scuse full chapters
71 pages
Ss Activity 6
No ratings yet
Ss Activity 6
2 pages
8x8 LED Matrix MAX7219 Tutorial With Scrolling Text & Android Control Via Bluetooth - HowToMechatronics
100% (1)
8x8 LED Matrix MAX7219 Tutorial With Scrolling Text & Android Control Via Bluetooth - HowToMechatronics
43 pages
1.what Is Cursor ? Answer
No ratings yet
1.what Is Cursor ? Answer
7 pages
For Language Basics: C++ Placement Course
No ratings yet
For Language Basics: C++ Placement Course
2 pages
CSE 390 Bash Command Reference
100% (1)
CSE 390 Bash Command Reference
3 pages
Da Programs
No ratings yet
Da Programs
10 pages
Seminar On NUMERICAL RELAY
No ratings yet
Seminar On NUMERICAL RELAY
18 pages
Aermod Userguide PDF
No ratings yet
Aermod Userguide PDF
321 pages
AM4 Sample Test
No ratings yet
AM4 Sample Test
4 pages
Embedded System Technology 2008 13
No ratings yet
Embedded System Technology 2008 13
34 pages
Instructions:: IT185L - (CISCO 2) Actual Lab Skills Integration Challenge
0% (1)
Instructions:: IT185L - (CISCO 2) Actual Lab Skills Integration Challenge
3 pages
Aisc Shapes Database v15.0
No ratings yet
Aisc Shapes Database v15.0
15 pages
CGS Civil 3D Tools - 2016 - Brochure - ENG PDF
No ratings yet
CGS Civil 3D Tools - 2016 - Brochure - ENG PDF
4 pages
RFID-S Startup Guide
No ratings yet
RFID-S Startup Guide
12 pages
The Future of Robots - Rodney Brooks
No ratings yet
The Future of Robots - Rodney Brooks
2 pages
Data Filtering in Pandas
No ratings yet
Data Filtering in Pandas
23 pages
Homework 10: Section 4.2
No ratings yet
Homework 10: Section 4.2
8 pages
Unit 2 - QUEUE
No ratings yet
Unit 2 - QUEUE
30 pages
Carlos Rodríguez: Supply Chain Manager
No ratings yet
Carlos Rodríguez: Supply Chain Manager
3 pages
Manual LAMMPS
No ratings yet
Manual LAMMPS
1,175 pages