Technical Seminar Fuzzy Based Framework

VISVESVARAYA TECHNOLOGICAL UNIVERSITY
Jnana Sangama, Santhibastawad Road, Machhe

Belagavi - 590018, Karnataka, India
TECHNICAL SEMINAR (15ISS86) REPORT

ON
Feature Based Fuzzy Framework for Sentimental Analysis of Web Data
Submitted in the partial fulfilment of the requirements for the award of the degree of
BACHELOR OF ENGINEERING
IN
INFORMATION SCIENCE AND ENGINEERING
For the Academic Year 2019-2020
Submitted by
T S Thanya Gowda 1JS16IS083
Under the Guidance of
Mrs. Sahana V
Assistant Professor,
Dept. of ISE, JSSATEB
2019-2020
DEPARTMENT OF INFORMATION SCIENCE AND ENGINEERING
JSS ACADEMY OF TECHNICAL EDUCATION
JSS Campus, Dr. Vishnuvardhan Road, Bengaluru-560060
JSS MAHAVIDYAPEETHA, MYSURU
JSS ACADEMY OF TECHNICAL EDUCATION
JSS Campus, Dr. Vishnuvardhan Road, Bengaluru-560060
DEPARTMENT OF INFORMATION SCIENCE & ENGINEERING
CERTIFICATE
This is to certify that Technical Seminar (15CSS86) Report entitled “Feature Based
Fuzzy Framework for Sentimental Analysis of Web Data” is a bonafide
work carried out by T S Thanya Gowda in partial fulfilment for the award of degree of
Bachelor of Engineering in Information Science and Engineering of Vishvesvaraya
Technological University Belagavi during the year 2019-2020.
Signature of the Internal Guide Signature of the HOD
Mrs. Sahana V Dr. Dayananda P

Assistant Professor, Prof. & Head,
Dept. of ISE, Dept. of ISE,
JSSATE, Bengaluru. JSSATE, Bengaluru.
1.
2.
Signature of External Examiners & Internal Examiners

ACKNOWLEDGEMENT
The satisfaction and euphoria that accompany the successful completion

of any task would be incomplete without the mention of the people who made it
possible. So, with gratitude, we acknowledge all those whose guidance and
encouragement crowned my effort with success.
First and foremost, we would like to thank his Holiness Jagadguru Sri
Shivarathri Deshikendra Mahaswamiji and Dr. Mrityunjaya V Latte,
Principal, JSSATE, Bangalore for providing an opportunity to carry out the
Internship (15IS84) as a part of our curriculum in the partial fulfilment of the
degree course.
We express our sincere gratitude for our beloved Professor & Head of the
department, Dr. Dayananda P, for his co-operation and encouragement at all
the moments of our approach.
It is our pleasant duty to place on record our deepest sense of gratitude to

our respected guide Mrs. Sahana V, Assistant Professor, Department of ISE,
JSSATEB for the constant encouragement, valuable help and assistance in
every possible way.
We would like to thank all Teaching and non-teaching Faculty for

providing us with their valuable guidance and for being there at all stages of our
work.
T S Thanya Gowda
1JS16IS083
TABLE OF CONTENTS
1. INTRODUCTION 1
2. LITERATURE SURVEY 6
3. WEB DATA MINING 9
4. LEVEL OF SENTIMENT ANALYSIS 13
5. SENTIMENT CLASSIFICATION 15
6. OPINION LEXICONS 16
6.1 WORDNET GLOVES AND SENTIWORDNET 17
7. FUZZY BASED FRAMEWORK 18
8. APPLICATION OF SENTIMENT ANALYSIS 23
9. CONCLUSION 26
10. REFERENCES 27
ABSTRACT
Sentiment Analysis is an ongoing field of research in text mining field. Sentimental Analysis
is the computational treatment of opinions, sentiments and subjectivity of text. Many recently
proposed algorithms enhancements and various Sentimental Analysis applications are
investigated and presented briefly in this survey. These articles are categorized according to
their contributions in the various Sentimental Analysis techniques. The related fields to
Sentimental Analysis (transfer learning, emotion detection, and building resources) that
attracted researchers recently are discussed. The main target of this survey is to give nearly
full image of Sentimental Analysis techniques and the related fields with brief details. The
main contributions include the sophisticated categorizations of a large number of recent
articles and the illustration of the recent trend of research in the sentiment analysis and its
related areas.Sentiment Analysis is the use of natural language processing, text analysis,
computational linguistics, and biometrics to systematically identify, extract, quantify, and
study affective states and subjective information. With the advent of the web, people not only
consume content by downloading data, they also contribute data and produce new content.
They share their opinions on the web. Sentiment analysis and opinion mining branch out
from data science and natural language processing. Growth in the area of opinion mining
and sentiment analysis has been rapid and aims to explore the opinions or text present on
different platforms of social media through machine-learning techniques with sentiment,
subjectivity analysis or polarity calculations.
1. INTRODUCTION
Development in web technology and their rapid advances have produced path for generation
of vast size of data in the web for internet users. Sentimental analysis refers to scrutinizing of
online web reviews in a precise and organized way. Opinion mining refers to an approach of
extracting people’s attitude related to a particular subject from a vast group of opinions or
reviews which are openly presented in web. Thus this mining or analysis is most needed. For
instance a client or user wants to buy a good, get details from people in sense of their
opinions. It’s a difficult task to derive a decision from these huge available reviews in web.
This demand for an automated system to mine the goodness and badness of a product which
will be effective and efficient for the web users for performing the decisions taking work at
online buying and selling, which is achieved using fuzzy based framework. The growth of the
web and information on social media, made sentiment analysis a relevant field to find the
opinions of others. Sentiment Analysis and opinion mining are two terms that are often used
interchangeably. Sentiment analysis search for the sentiment words/expression in a text and
then analyses it. Sentiment analysis (also known as opinion mining or emotion AI) refers to
the use of natural language processing, text analysis, computational linguistics,
and biometrics to systematically identify, extract, quantify, and study affective states and
subjective information. Sentiment analysis is widely applied to the voice of customers, this
includes reviews and survey responses, online and social media, healthcare materials for
applications ranging from marketing to customer service to clinical medicine. Whereas,
opinion mining extracts and analyses people's opinion about an entity.
Sentiment analysis is divided into two categories, Lexicon Analysis and Machine Learning.
Lexicon Analysis is used to calculate the polarity of documents from semantic orientation of
words or phrases in the documents. Machine Learning involves building models derived from
labelled training data. This training data includes sentences or instances of texts. The goal is
to find the document orientation.Sentiment Analysis can be Unimodal or Multimodal
sentiment analysis. Unimodal Sentiment Analysis focuses only on the aspect of “Text”.
Multimodal Sentiment Analysis involves the fusion of facial expression, text and
paralinguistic features. Multimodal Sentiment Analysis is still in the infancy stage and will
soon be a popular research area in the field of Sentiment Analysis. Opinion mining research
considers the computational treatment of subjective information contained in text. With the
rapid growth of available subjective text on the internet in the form of product reviews, blog
1
posts and comments in discussion forums, opinion mining can assist in a number of potential
applications in areas such as search engines, recommender systems and market research. One
approach for detecting sentiment in text present in literature concerns the use of lexical
resources such as a dictionary of opinionated terms. SentiWordNet [6] is one such resource,
containing opinion information on terms extracted from the WordNet database and made
publicly available for research purposes. SentiWordNet is built via a semi supervised method
and could be a valuable resource for performing opinion mining tasks: it provides a readily
available database of term sentiment information for the English language, and could be used
as a replacement to the process of manually deriving ad-hoc opinion lexicons. In addition,
SentiWordNet is built upon a semi automated process, and could easily be updated for future
versions of WordNet, and for other languages where similar lexicons are available. Thus, an
interesting research question is to assess how effective is SentiWordNet in the task of
detecting sentiment in comparison to other methods, and what are the potential advantages
that could be obtained from this approach. This paper proposes a method for applying
SentiWordNet to derive a data set of document metrics and other relevant features, and
performs an experiment on sentiment classification of film reviews using the polarity data set
introduced in [14]. We present and discuss the results obtained in light of similar research
performed using manually built lexicons, and investigate possible sources of inaccuracies
with this method. Further analysis of the results revealed opportunities for improvements to
this approach, which are presented in our concluding remarks.
2
Reviews dataset
 The above figure is the overall representation and flow of sentiment Analysis.
Based on the above figure, sentiment analysis can be divided into three different levels,
3
 Document Level Analysis: The task at this level is to determine the overall opinion of
the document. Sentiment analysis at document level assumes that
each document expresses opinions on a single entity.
 Sentence Level Analysis: Sentence level analysis is used to study each sentence to
determine whether they have expressed an opinion. If an opinion is expressed to
determine whether the opinion is positive, negative or neutral.
 Aspect Level Analysis: It is the technique in which the topics (aspects) are extracted
from the text using various algorithms and then sentiment for each topic is evaluated.
It can also be known as entity/feature-level sentiment analysis. Sentiments of multiple
entities present in a single sentence can be determined.
Fig. 3 The techniques of Sentiment Analysis

Sentiment Analysis techniques can be broadly classified into two categories,
Machine Learning Approaches: Machine Learning algorithms have been widely used for
sentiment analysis. Machine learning approaches are further divided into:
 Supervised Learning: Supervised learning is the machine learning task of learning a
function that maps an input to an output based on example input-output pairs. It infers
a function from labelled training data consisting of a set of training examples.
 Unsupervised Learning: Unsupervised learning is a type of machine
learning algorithm used to draw inferences from datasets consisting of input data
without labelled responses. The most common unsupervised learning method is
cluster analysis, which is used for exploratory data analysis to find hidden patterns or
grouping in data.
 Semi-supervised Learning: Semi-supervised learning is an approach to machine
learning that combines a small amount of labelled data with a large amount of
4
unlabelled data during training. Semi-supervised learning falls between unsupervised

learning (with no labelled training data) and supervised learning (with only labelled
training data).
Lexicon based approaches: Application of a lexicon is one of the two main approaches
to sentiment analysis and it involves calculating the sentiment from the semantic orientation
of word or phrases that occur in a text. It is further divided into:
 Dictionary Based: Dictionary-based sentiment analysis is a computational approach to
measuring the feeling that a text conveys to the reader. In the simplest
case, sentiment has a binary classification: positive or negative, but it can be extended
to multiple dimensions such as fear, sadness, anger, joy, etc.
 Corpus Based: Corpus based suggests data-driven approach where you will have
access not only to sentiment labels, but to a context which you can use to your
advantage in a machine learning algorithm.
5
2. LITERATURE SURVEY
Sentiment analysis and opinion mining are terms that refer to the field of study that analyses
opinions, evaluations, appraisals, attitudes and emotions of people towards entities such as
products, services, organizations, individuals, issues, events, topics and their attributes These
terms were used interchangeably to define opinions that entail positive or negative sentiments
Sentiment determines the polarity of opinions expressed in a given review. Cambria et al.
disputed the interchange of these concepts by classifying opinion mining as polarity detection
and sentiment analysis as focusing on emotion recognition. The opinion mining system only
needs to understand polarity that can be positive, negative or neutral sentiments depending on
the nature of sentences expressed in a review The process of detecting polarity is strongly
linked to analyzing sentiments on a particular subject. Most researches on sentiment analysis
are focused on descriptive data. Manke and Shivale explored the significance of social
networks as preferred environments for opinion mining and sentiment analysis. They
introduced the original method of opinion classification and tested their algorithm on real
social network datasets. They concluded from their findings that social networks exhibit
properties that make them suitable for opinion mining activities. Comprehensive surveys
have been presented on various methods used in opinion mining with limited focus on aspect
oriented analysis. The majority of current methods of sentiment analysis attempt to detect the
polarity of a review regardless of the entities such as hotels and facilities with their respective
aspects such as food and internet access for instance. By contrast, the task of this study is
concerned with aspect based sentiment analysis with the goal of identifying the aspects of
given target entities and sentiment expressed towards each aspect. The aspect based
sentiment analysis summarizes what people like and dislike from reviews of products or
services. It has always been a difficult task because several subtasks such as feature
extraction, feature grouping, polarity classification and evaluation measures have to be
performed to get an unbiased opinion, usually under the assumption of grammar free errors
which is not always realistic.
Sentiment classification concerns the use of automatic methods for predicting the orientation
of subjective content on text documents, with applications on a number of areas including
recommender and advertising systems, customer intelligence and information retrieval.
SentiWordNet is an opinion lexicon derived from the WordNet database where each term is
associated with numerical scores indicating positive and negative sentiment information. This
6
research presents the results of applying the SentiWordNet lexical resource to the problem of
automatic sentiment classification of film reviews. Our approach comprises counting positive
andnegative term scores to determine sentiment orientation, and an improvement is presented
by building a data set of relevant features using SentiWordNet as source, and applied to a
machine learning classifier. We find that results obtained with SentiWordNet are in line with
similar approaches using manual lexicons seen in the literature. In addition, our feature set
approach yielded improvements over the baseline term counting method. The results indicate
SentiWordNet could be used as an important resource for sentiment classification tasks.
Additional considerations are made on possible further improvements to the method and its
use in conjunction with other techniques.
The analysis of sentiment can be performed using two basic approaches of
1. Lexicon-based :
The lexicon-based approaches are dependent on the obtainability of a sentiment lexicon [7].
A sentiment lexicon is a group of previously created and known sentiment words. These
approaches could be classified into two dissimilar sets: (I) Dictionary based, which is a
computational approach to measuring the feeling that a text conveys to the reader.
Dictionaries for lexicon-based approaches can be created manually. In Dictionary-based
approach, firstly the opinion word from review text are found, which is followed by finding
their synonyms and antonyms from dictionary. The dictionaries like WordNet,
SentiWordNet, SenticNet may be incorporated for mapping and scoring. (II) corpus-based,
which uses semantic methods or statistical methods to search sentiment polarity [7]. Corpus-
based method helps to find opinion word in a context specific orientation. Beginning with a
list of opinion word, the corpus-based approach finds other opinion word in a huge corpus. A
hybrid approach combining the machine learning and the dictionary-based approaches may
be used for sentiment analysis. It employs the lexicon-based approach for sentiment scoring
followed by training a classifier assign polarity to the entities in the newly find reviews.
Hybrid approach is generally used since it achieves the best of both worlds, high accuracy
from a powerful supervised learning algorithm and stability from lexicon-based approaches.
2. Machine learning:
The machine learning approach which this study is based upon utilizes two main learning
techniques of supervised and unsupervised learning. Supervised learning algorithms such as
the support vector machine, Naïve Bayes , K-nearest neighbor and convolutional neural
network deep learning have been applied for sentiment analysis. Supervised learning
7
Algorithm ms require the training of machine using labeled dataset. However, unsupervised
learning algorithms such as the K-means and Fuzzy C-means clustering do not require
training datasets because they learn by observation. The application of supervised learning
algorithms as applied in this study to volumes of labeled training data on hotel reviews can
provide insightful information that would help hotels imp rove performance and overall
ratings amongst competitors. Different evaluation measures such as true positive rate, false
positive rate, precision, recall, and F measure rate, receiver operating characteristic (ROC)
area and precision recall curve (PRC) area are the benchmark metrics used to determine the
accuracy of different supervised learning algorithms. However, to get accurate results of
these measures, there is the need to design a framework that fosters the automatic creation of
properly labeled datasets containing actual sentiments expressed by customers.
The success of sentiment analysis and opinion mining on the use of tools for executing
different NLP tasks. Tools that can be used for the NLP tasks include Red Opal which is
used for enabling users to find products based on aspects. Tools that are used to help
companies extract and analyse opinions of customers on products from blogs include
SenticNet, Luminoso, Factiva, Attensity and Converseon. The NLTK, OpenNLP and
Stanford CoreNLP are widely used NLP toolkits to support the implementation of basic NLP
tasks such as POS tagging, named entity recognition and parsing [11]. The W EKA system
isa widely used tool that contains a collection of techniques for data analysis and predictive
modelling. It supports several standard data mining tasks that include data pre-processing,
8
3. Web Data Mining Systems
Web data mining can be defined as the discovery and analysis of useful and relevant
information from the World Wide Web data. WWW has a lot of useful or useless Meta data,
web log data of users who retrieved multiple web pages and the structured and unstructured
data. Over the past Fifteen years, we have already faced lot of explosion type of information
resources available over the web. Web mining can be applied to all the fields of artificial
intelligence system, human interaction, cloud computing, neural data mining, geographical
data mining, and information retrieval etc. Many Web applications focused on extraction of
knowledge from the web, extraction of knowledge from the user’s behavior, getting
information from the web, providing information to the web, downloading and uploading
data over the web were developed. The main objective of the web mining is to provide data
mining algorithms which can improve the content, structure, usage, performance, and
categorization of web documents, snippets and user sessions. The Web data mining can be
classified into three categories namely Web Structure Mining, Web Content mining and Web
usage mining. All these three categories focus on the process of discovering unknown data
and potentially very useful information from the web. Though each of them focuses on the
same attribute each may be using it with different mining objectives on the web.
Fig 1: Web Mining Systems
9
Web Structure Mining

Web Structure Mining involves mining the structure of web document’s and links. Useful
insights can be given by mining the structural information on the web. WSM is very
useful in generating information such as visible web documents, luminous web documents
and luminous paths, a path common to most of the results returned, linkage information
useful to improve search engine results, hyperlink structure analysis, link analysis, graph,
categorization and mining the document structure.
Web Content Mining
Web Content mining examines the contents of web pages as well as the results of web
searches. WCM is described as the automatic search of information resources available on-
line. It represents structured, unstructured, semi structured documents and builds model for
interactive retrieval view and Data Base View. It is all about extracting and integrating of
useful data with the objective of information and knowledge discovery from Web page
contents. WCM has two different major approaches. One is Agent based approach and
another one is Data base based approach. First approach is on improving the information
finding and filtering which are carried out using intelligent search agents, or personalized
web agents. Second approach is on modeling the data on the web into a more structured form
by connecting with multilevel data bases and web query systems.
Web Usage Mining
Web Usage Mining focuses on several techniques that could help in learning or predicting
user behavior and navigation pattern of users using the web round the clock. It includes
the data from server access logs, user registration or profiles, user sessions or transactions
etc., It also depends on the collaboration of the user to allow the access of the web
log records.
Information Retrieval Systems
IR involves retrieving desired information from textual data. The historical development of
IR was based on effective use of libraries. Many universities and public libraries use IR
systems to provide access to books, journals and other documents. IR effectiveness can be
measured by using a technique called test collection. The test collection consists of document
collection, a test suite of information need express as queries and a set of relevant judgments,
standard binary assessment of either relevant or non-relevant response for each query
document pair. Automated IR systems are used to reduce information overload. Web search
engines are the most visible IR applications. In the work of searching, retrieving data from
10
web, we are using several other words such as data retrieval, document retrieval, information
retrieval, text retrieval. There a number of words overlapping but with similar meaning and
tasking almost similar. For the information retrieval to be efficient, the documents are
typically transformed into a suitable representation. There are several representation available
like theoretical, probabilistic and future based retrieval models. As a result, some traditional
data mining methods are not applicable to web mining data retrieval. The key problem of
information retrieval in web mining is how to improve comprehensive and correlated
information accessed from the web data base and make efficient information retrieval for
classification of different web pages for retrieving relevant information. The challenges for
IR is to deal with the structure of the hyperlinks within the web itself. IR Link analysis is an
old area of research. However with the growing interest in web mining, the research on
structure analysis has increased and this had resulted in a new emerging research called link
mining which will help in retrieving accurate information without spreading spam, even from
irrelevant or unwanted web pages.
Web Log Data Preprocessing
Web Usage Mining is one of the applications of data mining techniques to discover usage
patterns from global Web data. WUM includes data preprocessing, pattern discovery and
pattern analysis. In the preprocessing phase raw Web logs are cleaned, analyzed and then
converted in to pattern mining process. Data pre-processing includes the following process:
cleaning, normalization, transformation, feature extraction and selection etc., of the data
recorded in the server logs system. The logs are used to identify users and sessions.
Sometimes server logs analysis is not accurate and reliable. So the system also considers
cookies and sessions. The server logs authentication of server logs must be formalized with
standard set of format and it should be updated to capture user access data. Most of the
preprocessing techniques suffer from low quality. So the system should improve the quality
of preprocessed data and their algorithms. A new technique is essential to analyze the log file.
The Basic Process of web Log Mining should concentrate in Data Preparation, Data Mining,
and Pattern Analysis. Using these techniques, the problem can be solved and ultimately data
can be converted into knowledge. By doing a survey of literature on web log preprocessing, it
was found that web log systems have these essential techniques to get exact pattern. The steps
are data cleaning, data filtering, path completion, user identification, and session
identification, cluster of web session and data visualization. WLM is used to enhance server
performance, improve web site navigation, improve the design of web applications, and
improve the multidimensional web log analysis, identifying web access association or pattern
11
analysis. The pattern analysis is used to analyze web cashing, pre-fetching, swapping and
frequently used predefined reports. The report should include the following information:
HITS information, list of top requested URLs, referred, list of common browsers, HIT per
time and error report.
Web Sentiment Analysis
Sentiment mining or Opinion mining is the field of study that analyzes people’s opinions,
sentiments, evaluations, appraisals, and emotions towards entities such as products, services,
organizations, individuals, issues, events, topics and their attributes. In conversation mining
the additional task is to determine the nature of opinion whether it is positive or neutral in
general. Opinion mining is a type of Natural language processing for tracking the attitudes,
feelings or appraisals of the public about particular topic, product or services. Sentiment
analysis can be useful in several ways. For example, it tracks and judges the success rate of
an advertisement campaign or launch of new product, determines popularity of products and
services with its versions and also tells us about demographics which like or dislike particular
features. The analyst company gets a much clearer picture of public opinion than surveys or
focus groups, if this kind of information is identified in a systematic way.
12
4. Level of Sentiment Analysis

The sentiment analysis tasks can be accomplished at the following levels of granularity,
namely word, sentence, document, and feature level.
Fig 2: Level of Sentiment Analysis

Word Level Sentiment Analysis
The word level sentiment analysis is used to find semantic orientation at phrase level. Most
previous works use the prior polarity of words and phrases for sentiment classification at
sentence and document levels. There are two methods of automatically annotating sentiment
at the word level such as dictionary-based and corpus-based ones. The Dictionary based
sentiment analysis means a small seed list of words with known prior polarity is created. This
seed list is then extended by extracting synonyms or antonyms iteratively from online
dictionary sources like Word Net. Corpus based methods rely on syntactic or statistical
techniques like co-occurrence of word with another word whose polarity is known.
Sentence Level Sentiment analysis
At sentence level sentiment analysis is used to detect subjective sentences in a document
from a mixture of objective and subjective sentences and also the sentiment orientation of
these subjective sentences is determined.
Document Level Sentiment Analysis
13
Document-level sentiment analysis is considering the whole document as the basic unit
whose sentiment orientation is to be determined. To simplify the task, it is presumed that
each text‘s overall opinion of each text is completely held by a single opinion holder and is
about a single object.
Features Based Sentiment Analysis
This level of sentiment analysis is used to extract system feature and the corresponding
opinion about it. The opinion may be positive or negative of a particular system. The person
may like some features and dislike some, even though the general opinion of the system may
be positive or negative.
14
5. Sentiment Classification
Sentiment classification is an opinion mining activity concerned with determining what, if
any, is the overall sentiment orientation of the opinions contained within a given document. It
is assumed in general that the document being inspected contains subjective information,
such as in product reviews and feedback forms. Opinion orientation can be classified as
belonging to opposing positive or negative polarities – positive or negative feedback about a
product, favorable or unfavorable opinions on a topic – or ranked according to a spectrum of
possible opinions, for example on film reviews with feedback ranging from one to five stars.
Supervised learning methods using different aspects of text as sources of features have been
proposed in the literature. Early work seen in [13] presents several supervised learning
algorithms using bag-of words features common in text mining research, with best
performance obtained using support vector machines in combination with unigrams.
Classifying terms from a document into its grammatical roles, or parts of speech has also
been explored: In [21] part of speech information is used as part of a feature set for
performing sentiment classification on a data set of newswire articles, with similar
approaches attempted in [10], [7] and [16], on different data sets. On [20] a method that
detects and scores patterns in part of speech is applied to derive features for sentiment
classification, with a similar idea applied to opinion extraction for product features seen in
[4]. Separation of subjective and objective sentences for the purposes of improving document
level sentiment classification are found in [14], where considerable improvements were
obtained over a baseline word vector classifier. Other studies focus on the correlation of
writing style to overall sentiment, taking into account the use of colloquialisms and
punctuation that may convey sentiment. In [22] a lexicon of colloquial expressions and a
regular expression rule base is created to detect unique opinion terms such as unusual
spellings (“greeeat”) and word combinations (“supergood”). In [1] document statistics and
features measuring aspects of writing style are combined with word vectors to obtain
considerable improvements over a baseline classifier on a data set of film reviews.
15
6. Opinion Lexicons
Opinion lexicons are resources that associate sentiment orientation and words. Their use in
opinion mining research stems from the hypothesis that individual words can be considered
as a unit of opinion information, and therefore may provide clues to document sentiment and
subjectivity.
Manually created opinion lexicons were applied to sentiment classification as seen in [13],
where a prediction of document polarity is given by counting positive and negative terms. A
similar approach is presented in the work of Kennedy and Inkpen [10], this time using an
opinion lexicon based on the combination of other existing resources. Manually built lexicons
however tend to be constrained to a small number of terms. By its nature, building manual
lists is a time consuming effort, and may be subject to annotator bias. To overcome these
issues lexical induction approaches have been proposed in the literature with a view to extend
the size of opinion lexicons from a core set of seed terms, either by exploring term
relationships, or by evaluating similarities in document corpora. Early work in this area seen
in [9] extends a list of positive and negative adjectives by evaluating conjunctive statements
in a document corpus. Another common approach is to derive opinion terms from the
WordNet database of terms and relationships [12], typically by examining the semantic
relationships of a term such as synonyms and antonyms. Lexicons built using this approach
can be seen applied to subjectivity detection research in [21] and applied to sentiment
classification in [4] and [16].
6.1 WordNet Glosses and SentiWordNet
As noted ,term relationships in the WordNet database form a highly disconnected graph, and
thus expansion of opinion information from a core of seed words by examining semantic
relationships such as synonyms and antonyms is bound to be restricted only to a subset of
terms. To overcome this problem, information contained in term glosses – explanatory text
accompanying each term – can be explored to infer term orientation, based on the assumption
that a given term and the terms contained in its gloss are likely to indicate the same polarity.
In [2] a method for lexicon expansion is proposed where terms are assigned positive or
negative opinions based on the existence of terms known to carry opinion content found on
the term gloss. The authors argue that glosses have a potentially low level of noise since they
“are designed to match as close as possible the components of meaning of the word, have
relatively standard style, grammar and syntactic structure”; This idea is also seen in [5], this
16
time by using supervised learning methods for extending a lexicon by exploring gloss
information, yielding positive accuracy improvements over a gold standard in comparison to
some of the methods previously discussed in this section. This is the same approach
employed on building the SentiWordNet opinion lexicon [6].
SentiWordNet is built in a two-stage approach: initially, WordNet term relationships such as
synonym, antonym and hyponymy are explored to extend a core of seed words used in [19],
and known a priori to carry positive or negative opinion bias. After a fixed number of
iterations, a subset of WordNet terms is obtained with either a positive or negative label.
These term’s glosses are then used to train a committee of machine learning classifiers. To
minimize bias, the classifiers are trained using different algorithms and different training set
sizes. The predictions from the classifier committee are then used to determine the sentiment
orientation of the remainder of terms in WordNet.
17
7. Fuzzy Based Framework

An effective fuzzy based framework is proposed that deploys SentiWordNet to generate
textual review counts and labelling reviews into seven levels as strong-negative, negative,
weak-negative, neutral, weak-positive, positive, and strong-positive. The score values
generated the datasets of mobile and laptop reviews will be scrutinized among seven classes
of sentiment and categorized with various machine learning techniques which have ensued
with greater results.
An expansion in online activities like blogging, social networking, emailing, review posting,
and so forth has resulted in very vast gathering of user-generated content. In this, a fuzzy
based approach fine-level sentiment investigation of web reviews is done through
assimilating opinion descriptors with fuzzy linguistic hedges.
Fuzzy Logic is a Bendable machine learning methodology which imitates the human thinking
rationality. In general logic will represent binary values and it signifies probable binary
explanations. Fuzzy logic represents many values and it selects and defines an inbetween
value that will offer an implication tool which can be used for the task of command execution
and interpretation.
Proposed framework has three phases:

1. Pre-processing phase
2. Feature Selection Phase and
3. Fuzzy based Sentiment Analysis phase.
The Fig. 2 showcases the proposed feature based
18
Fig. 2. Proposed Fuzzy based Framework for Sentimental Analysis.
Pre-processing Phase
A. Data Acquisition:
Using JSoup Crawler a large amount of reviews from online social portals is gathered, which
parses every web page into only natural text pruning all other tag data. The obtained input
dataset is kept in a textual document way.
B. Data Pre-Processing
1. Stop words Removal:

Stop words indicate commonly used arguments in any English statements which need to be
ignored in processing. Stop words have to be cleaned out earlier or later to the dealing out in
19
natural language. At, a, is, the, an, which, on etc. are sample stop words which have to be
pruned out from the input text. Here around 3855 such words have been detached.
2. Stemming:
Web opinions normally are normally in the casual way those having huge variety of web
based flaky jargon embedded with in any of the needed comments. These may represent any
of the extra ing formats, or bunch of slang styled terms and so on. Such form of terms have to
be rescanned by the automated system and it has to be pruned at its stem level for retaining
only the root term of the slang word that can be a more filtered way of keeping the input
text ready for the fast processing of the future operations in the projected methodology.
Porter Stemming technique is deployed in this work. Following are the stages involved in
this process.
Stage 1: prune the all -ed, -ing and plurals

Stage 2: take it’s changed from i to y station
Stage 3: Plan for dual suffixes to solos.
Stage 4: handle all the suffixes with full and ness and so on.
Stage 5: get all –ant, -ences and so on.
Stage 6: Rejects the last ‘e’s.
Following are some examples showing process of stemmer.

• Accesses -access
• Repeats- repeat
• Predicate – predict
• Conclude - conclude
• Cumulating – cumulate
• Following - follow etc.
3. Parts Of Speech Tagging:

Here corresponding parts of speech are tagger with the input reviews where every input
sentence is parsed followed by the process where individual word is tagged with equation (1)
and (2) for obtaining its part of speech like Adjective (JJ), Personal Pronoun (PRP), Proper
Noun Singular (NNP) etc. Used tagger is designed with Maximum Entropy model that is as
same as the design of stochastic tagging along with Penn Treebank tag set.
20
The Table I list few POS tag with their meaning.
Feature Selection Phase

In Feature Selection Phase, feature set for sentimental analysis is generated using POS
tagging that generates POS tagged outputs which are given as input to SentiWordNet for
generating the Score class values.
Fuzzy Based Sentiment Analysis Phase

In the final phase of Fuzzy based Sentiment Analysis, a fine-level sentimental analysis online
review is done. The reviews are classified as strong-negative, negative, weak negative,
neutral, weak-positive, positive, and strong positive. Further a new web review is classified
using fuzzy sentiment score determined in the following steps:
a. Mine the connected descriptors, hedges and features as shown in feature Table II.
b. Ascertain starting value and SentiWordNet score based feature descriptors polarity
c. Determine complete score of sentiment by means of fuzzy rules.
d. In the final step, fine-level sentimental analysis of online reviews is performed
e. Once the Overall Fuzzy score is obtained, for effective classification the following rules
are applied: (SC = Sentimental Class) as shown in Table III.
21
TABLE III: FUZZY CLASSIFICATION RULES
22
8. Applications of Sentiment Analysis
Sentiment Analysis has a wide range of applications. It has a great impact on product sales
and business development. Customer sentiment analysis examines the emotions, impressions
and attitudes surrounding your business to make sales and marketing decisions. The
applications of sentiment analysis in business cannot be overlooked. Sentiment analysis in
business can prove a major breakthrough for the complete brand revitalization. The key to
running a successful business with the sentiments data is the ability to exploit the unstructured
data for actionable insights. Machine learning models, which largely depend on the manually
created features before classification, have served this purpose fine for the past few years.
Thanks to sentiment analysis, companies can understand the reputation of their brand. By
analysing social media posts, product reviews, customer feedback, or NPS responses (among
other sources of unstructured business data), they can be aware of how their
customers feel about their product. They can also track specific topics and get relevant
insights on how people are talking about those topics. Sentiment analysis is particularly
useful for social media monitoring because it goes beyond metrics that focus on the number
of likes or retweets, and provides a qualitative point of view. Let’s say a company has just
launched a new product feature and you notice a sharp increase in mentions on Twitter.
However, receiving tons of mentions does not necessarily mean a good thing. Are customers
tweeting more because they are expressing good things about this new product feature? Or,
are customers actually complaining about the feature having lots of bugs? Performing Twitter
sentiment analysis can be an excellent way to understand the tone of those mentions and
obtain real-time insights on how users are perceiving your new product. Twitter sentiment
analysis systems allow you to sort large sets of tweets and detect the polarity of each
statement automatically. And the best part, it’s fast and simple, saving teams valuable hours
and allowing them to focus on tasks where they can make a bigger impact. Almatarneh &
Gamallo (2018) studied the impact of extreme views toward the product sales [1]. Their
review indicates that with the increase in negative online customer reviews there will be an
increase in the customer’s negative attitude towards the products. At its most basic, sentiment
analysis is a social media analytics tool that involves checking how many negative and
positive keywords are present in a chunk of conversation. If there are more positive keywords
than negative, it is considered positive content. If there are more negative keywords, it is
called negative content. But there’s a lot more to it than that, and it’s real worth is found in
the details. In-depth analysis involves finding opinions in social media content and extracting
23
the sentiment they contain. An opinion is made up of a target, also called a topic, and a
sentiment on the topic. Stock market prediction has been identified as a very important
practical problem in the economic field. However, the timely prediction of the market is
generally regarded as one of the most challenging problems due to the stock market’s
characteristics of noise and volatility. To address these challenges, we propose a deep
learning-based stock market prediction model that considers investors’ emotional tendency.
First, we propose to involve investors’ sentiment for stock prediction, which can effectively
improve the model prediction accuracy. Second, the stock pricing sequence is a complex time
sequence with different scales of fluctuations, making the accurate prediction very
challenging. We propose to gradually decompose the complex sequence of stock price by
adopting empirical modal decomposition (EMD), which yields better prediction accuracy.
Third, we adopt LSTM due to its advantages of analysing relationships among time-series
data through its memory function. We further revised it by adopting attention mechanism to
focus more on the more critical information. Experiment results show that the revised LSTM
model can not only improve prediction accuracy, but also reduce time delay. It is confirmed
that investors’ emotional tendency is effective to improve the predicted results; the
introduction of EMD can improve the predictability of inventory sequences; and the attention
mechanism can help LSTM to efficiently extract specific information and current mission
objectives from the information ocean. Evaluation of the effectiveness of a stock price
prediction model was based on two dataset including historical price and mood information
dataset. [13]. The historical stock prices, stock prediction, the investor’s mood as well as the
discussion related to organization’s management were extracted from Yahoo Finance
message board based on 18 selected companies. The SVM was used as a classifier together
with the six features including price, user sentiment, sentiment classification, Latent Dirichlet
Allocation (LDA) based method, joint sentiment/topic (JST) based method and aspect-based
sentiment had been incorporated to evaluate the effectiveness of sentiment analysis in
predicting stock market movement. Yu & Wang (2015) presented sentiment analysis on fans
tweets during five selected games of FIFA World Cup 2014 [14]. The goal was to identify the
audience’s emotional responses during and after the games, especially when their own or
opponents’ players hit the goals. They concluded that fear and anger were dominant emotions
expressed by audiences when their opponent team won and that emotional state (fear and
anger) reduced when their chosen team won. Philander & Zhong (2016) analysed the
customer’s sentiment analysis towards Las Vegas resorts through Twitter data [15]. A
sentiment index is created using a sentiment lexicon methodology based on Twitter data and
24
the sentiment score is compared with the data from trip advisor. These results were used to
examine the external validity. It is shown that both sentiment metrics and trip advisor data are
very similar in terms of convergent and discriminant validity.
25
CONCLUSION
In real world, web analysis plays an important role in understanding the data and knowledge
discovery from the real input data. In web sentiment analysis, understanding the data and
knowledge discovery are the crucial parts of data mining process. In analyzing the data, the
relevant data are extracted and used for prediction in data mining. In relevant data extraction,
Sentiment prediction plays the role of identifying the highly relevant features from the
original data. Web sentiment mining is a new and promising research area which will to help
users in gaining insight into overwhelming information on the web social media. An effective
framework for sentimental analysis of web data is proposed using fuzzy based machine
learning algorithm to accomplish fine-level sentiment analysis. Proposed fuzzy based
approach is matched with various advanced approaches and is investigated on Mobiles and
Laptops reviews which demonstrate upper indication of accurateness through showcasing
lesser error rates. Thus proposed approach can be effectively used as a tool for performing
analysis of opinions from online reviews of any domain in the web. Thus it can serve for all
online users to take better and profitable decisions for them in their any online activities.
This work can be further enhanced by incorporating the audio texts available in online.
26
REFERENCES
[1] A. Kennedy and D. Inkpen, “Sentiment classification of movie reviews using contextual
valence shifters”, Computational Intelligence, vol. 22, no. 2, pp. 110–125, 2006.
[2] Ajit Danti, Shoiab Ahmed – “A Novel Approach for Sentimental Analysis and Opinion
Mining based on SentiWordNet using Web Data”, International Conference on Trends in
Automation, Communication & Computing Technologies, pp. 07-11, IEEE Xplore,
December 2015.
[3] Bi-Ru Dai and Po-Wei Liang, “Opinion Mining on Social Media Data”, IEEE 14 th
International Conference on Mobile Data Management, pp. 91–96, 0.1109/MDM.2013.73,
2013.
[4] Chen H, Dang Yan, Yulei Zhang and Hsinchun, A lexicon enhanced method for
sentiment classification: An experiment on online product reviews. IEEE Intelligent Systems,
25(4):46–53, 2010.
[5] Dave K., Lawrence S., Pennock D. M., “Mining the Peanut Gallery: Opinion Extraction
and Semantic Classification of Product Reviews”, in Proceedings of the 12th International
Conference on World Wide Web, ACM, pp- 519–528, 2003.
[6] Emelda C, “A Comparative Study on Sentimental Classification and Ranking Reviews”,
Int. J. Innovative Res. Adv. Eng., ISSN 2349–2163, 1(10), 2014.
[7] Fei Liu and Gang Li, “A Clustering - based approach on sentiment analysis”,Conference
on Intelligent System and Knowledge Engineering (ISKE), pp. 331-337, ISKE.2010.
5680859, Hangzhou, China, 2010.
[8] G. Anuradha and D Joel Varma, “Fuzzy Based Summarization of Product Reviews or
Better Analysis”, Indian Journal of Science and Technology, Vol 9(31), August 2016.
[9] L Polanyi and A Zaenen, “Contextual valence shifters,”in Computing Attitude and Affect
in Text: Theory and Applications”, vol. 20, The Information Retrieval Series, pp. 1–10, 2006.
[10] Lianghao Li et al, “Multi-domain Active Learning for text classification”, Proceedings
of the 18th ACM SIGKDD International Conference on Knowledge discovery and data
mining, ACM, 2012.
[11] Liu B., “Sentiment Analysis and Opinion Mining, Synthesis Lectures on Human
Language Technologies”, San Rafael, Calif, Morgan & Claypool, 5(1): pp-1–167, 2012.
[12] M Trupthi, Suresh Pabboju, G. Narasimha, “Sentiment Analysis on Twitter Using
Streaming API”, Advance Computing Conference (IACC), 2017 IEEE 7 th International,
ISSN: 2473-3571, Hyderabad, IEEE, Jan- 2017.
[13] Mita K Dalal and Mukesh A Zaveri, “Opinion Mining from Online User Reviews Using
Fuzzy Linguistic Hedges”. In Applied Computational Intelligence and Soft Computing,
Volume 2014, Article ID 735942, 9 pages http://dx.doi.org/10.1155/2014/735942, 2014.
[14] Pang B., Lee L., Vaithyanathan S., “Thumbs up?: Sentiment Classification using
Machine Learning Techniques”, in Proceedings of the ACL-02 Conference on Empirical
Methods in Natural Language Processing- Volume 10, Association for Computational
Linguistics, pp-79–86, 2002.
[15] Shoiab Ahmed, Ajit Danti –“Effective Sentimental Analysis and Opinion Mining of
Web Reviews Using Rule Based Classifiers”; international conference on Computational
Intelligence in Data Mining, Volume1, Pages171-179, SPRINGER publications, December
2015.
27

Technical Seminar Fuzzy Based Framework

Uploaded by

Copyright:

Available Formats

Technical Seminar Fuzzy Based Framework

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Technical Seminar Fuzzy Based Framework

Uploaded by

Copyright:

Available Formats

VISVESVARAYA TECHNOLOGICAL UNIVERSITY

Jnana Sangama, Santhibastawad Road, Machhe

TECHNICAL SEMINAR (15ISS86) REPORT

T S Thanya Gowda 1JS16IS083

Under the Guidance of

DEPARTMENT OF INFORMATION SCIENCE & ENGINEERING

Signature of the Internal Guide Signature of the HOD

Mrs. Sahana V Dr. Dayananda P

Signature of External Examiners & Internal Examiners

The satisfaction and euphoria that accompany the successful completion

It is our pleasant duty to place on record our deepest sense of gratitude to

We would like to thank all Teaching and non-teaching Faculty for

3. WEB DATA MINING 9

4. LEVEL OF SENTIMENT ANALYSIS 13

6.1 WORDNET GLOVES AND SENTIWORDNET 17

7. FUZZY BASED FRAMEWORK 18

8. APPLICATION OF SENTIMENT ANALYSIS 23

Fig. 3 The techniques of Sentiment Analysis

unlabelled data during training. Semi-supervised learning falls between unsupervised

3. Web Data Mining Systems

Fig 1: Web Mining Systems

Web Structure Mining

4. Level of Sentiment Analysis

Fig 2: Level of Sentiment Analysis

7. Fuzzy Based Framework

Proposed framework has three phases:

The Fig. 2 showcases the proposed feature based

Fig. 2. Proposed Fuzzy based Framework for Sentimental Analysis.

1. Stop words Removal:

Stage 1: prune the all -ed, -ing and plurals

Following are some examples showing process of stemmer.

3. Parts Of Speech Tagging:

The Table I list few POS tag with their meaning.

Feature Selection Phase

Fuzzy Based Sentiment Analysis Phase

TABLE III: FUZZY CLASSIFICATION RULES

8. Applications of Sentiment Analysis

You might also like