International Journal of Advances in Engineering and Management (IJAEM)

Volume 4, Issue 5 May 2022, pp: 1307-1311 www.ijaem.net ISSN: 2395-5252

Sentiment Analysis on Product Reviews

using Web Scrapping
(Kajal Sharma, Harsh Sharma)1, Er. Ankit Khare2
1UG student of Department of Information Technology, Shri Ramswaroop Memorial College of Engineering
and Management Lucknow, Uttar Pradesh, India
2Assistant Professor, Department of Information Technology, Shri Ramswaroop Memorial College of
Engineering and Management Lucknow, Uttar Pradesh, India

--------------------------------------------------------------------------------------------------------------------------------------
Submitted: 10-05-2022 Revised: 19-05-2022 Accepted: 22-05-2022
-----------------------------------------------------------------------------------------------------------------------------------
ABSTRACT – Nowadays the rise in demand for e- of information within the variety of reviews is
commerce with people preferring online purchasing produced which helps prospective buyers to decide
of products, even we’ve got seen the rise in on the proper product. Furthermore these reviews
purchasing products online in these pandemic containing opinion of other customers about the
situation. There’s an unlimited amount of data products which might be useful for the corporate to
being shared. The e-commerce websites are loaded spot the areas which require to be enhanced within
with large volume of knowledge. Even social the products. Besides the reviews and ratings
media has become the nice help for selling and provided do little to assess the particular features of
buying goods, and an excellent deal in sharing of the products. Thanks to all the Constrains the
this information throughout the world.This has customer is unable to form a call about the product.
influenced consumer’s habits everywhere the Sentiment Analysis is also known as
world. Thanks to vivid reviews provided by the opinion mining can be used to extract customer
shoppers, there’s a feedback environment being reviews from different sources on the internet. This
developed in e-commerce websites for helping technique implements various algorithms to
customers buy the correct products and guiding the analyze the corpus of data and make sense out of it.
businesses to boost the features of products suiting It helps to identify the orientation of a sentences
consumer’s demand. The sole disadvantages for the thereby recognizing the element of positivity or
purchasers is availability of this huge volume of negativity in it. Opinion mining can be
data and its non-uniformness structure. The implemented through a web scrapping based
customers find it difficult to exactly find the review approach. Opinion mining uses natural language
for particular features of a product that they processing to extract the subjective information
intended to buy. Also there are lots of review that from the data (in this case its customer reviews for
they find difficult to read all and make decision for the product).
the product. We propose a technique to find the
exact opinion about product and further classifying 1.1 PROBLEM STATEMENT
it into positive and negative reviews using web A web application that collects reviews
scrapping based approach that help customers to from particular e-commerce website where
make decision for buying that product or not. customer want to buy a certain product and
analyzes them. It would segregate the reviews into
Key Words: Sentiment Analysis, Opinion Mining, positive and negative reviews. The reviews will be
Web Scrapping, Product Reviews, Natural helpful to the customers to take decision for buying
Processing Language etc. the product based on the user’s feedbacks. The
application further provides the pros and cons of
I. INTRODUCTION the individual feature of the product and also
Recently we’ve witness the rise of provide reports about sentiment analysis performed
shopping for products or goods online and within on the products. We further aim to create a
the recent years E-Commerce websites has recommendation system that recommends products
explored everywhere within the world. Majority of to users according to the feature requirement of
population is preferring to buy products through users.
these websites online. Therefore the great amount

DOI: 10.35629/5252-040513071311 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1307
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 5 May 2022, pp: 1307-1311 www.ijaem.net ISSN: 2395-5252

1.2 MOTIVATION review document is predicted by sentimental

Averyusefulfor recognizing the opinion of score. [4]
any particular product. Regardless of the name, the  Sentiment Analysis and Prediction of Online
real motivation of sentiment analysis is the same, to Reviews with Empty Ratings [2018] : In this
know a user or audience opinion on a target object paper,Sasikal, L.Mary Immaculate Sheela used
by analyzing a vast amount of text from various Naïve bayes algorithm and logistic regression,
sources. We are able to analyze text on different Multinomial and Bernoulli classifier for the
level of details, and the detail level depends on positive and negative reviews. [5]
customer’s opinion. Sentiment analysis is already
using in various organizations as Reputation III. TECHNOLOGIES
Management, Social –Media monitoring, Brand  Python: Python is a widely used dynamic
Monitoring, Market Research, Competitor Analysis programming language with a clean syntax and
as well as Product Analysis.As I see in my an indentation structure easy to learn [6].
surroundings I came to know many people don’t  Streamlit:Streamlit is an open-source Python
like to read all those reviews, they are long or in library that makes it easy to create and share
large volume.They are unable to make decisions beautiful, custom web apps for machine
better so that they can use our Sentiment analyzing learning and data science.
by following only few steps to get an efficient  NLP:NLP is natural language processing tool.
analysis which they want. It has a comprehensive toolkit with a good
range of grammar checking tools. It is fast and
II. LITERATURESURVEY reliable. It identifies the part of speech of the
 Sentimental Analysis on online product review words in a sentence. It is flexible and
[2017]: In this paper, Raheesa Safrin, E.A. extensible
Vimal used feature vector classification for  BeautifulSoup: Beautiful Soup is a third party
analysis of sentiment. The data used in this Python library from Crummy designed for
study is online product review collected from scraping.
sample website that we have created word such
as an adjective, adverb are able to convey IV. METHODOLOGIES
opposite sentiment with thehelp of negative The dataset used for this project is the
prefix, negative phrase identification is used. Amazon Reviews Database [7]. The reviews in the
[1] dataset are consists of the attributes such as:
 Sentimental Analysis of Product review Reviewer ID, Product ID, Review Text, Rating and
[2014]: In this paper, AartiPatil used advanced time of the review. The main source of data used is
Naïve Bayesian Algorithm technique to find the product reviews from Amazon. The reviews for
whether the reviews are positive or negative a few popular phones have been obtained by
for analysis of sentiment. The sentiment building a web crawler. The web crawler has been
analysis is used to extract, aggregate and written in Python using a scraping library called
analyse the opinion on product from discussion BeautifulSoup. Along with the review text, some
forum. [2] additional data related to the reviews such as
 Comparative Analysis of Sentimental reviewer name, review date, overall rating and
Orientation using SVM and Naïve Bayes comments were also obtained. The crawler is called
Techniques (2016): In this paper, ShwetaRana periodically to get themost up-to-date reviews.
used Naïve Bayes classifier in system. Each review is generally treated as a sentence or a
Sentiment classified as positive or negative group of sentences. They are cleaned and stored in
sentiment using film user review. Algorithm a CSV file.
like Naïve Bayes, Linear SVM and Synthetic Then, sentiment analysis is performed on
words is used. [3] the preprocessed reviews and overall sentiment
 Extracting a sentiment from Review: A score for each review is generated. Further for
Lexicon Based Approach [2017]: In this paper, feature extraction, there are two cases:
SujutaSonavane used Lexicon Based Approach 1. Single Feature: If the review contains
for extracting sentiment such as positive, only a single feature, then the sentiment score of
negative or neutral. Extracting the useful the review is assigned to the feature.
content from the opinion sources becomes a 2. Multiple Features: Some reviews have
challenging task. To extract sentiment from multiple features contained in them. So the above
review SentiWordNet is used to assign the procedure will not work in this case. Rules are
polarity of sentiment. The classification of
DOI: 10.35629/5252-040513071311 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1308
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 5 May 2022, pp: 1307-1311 www.ijaem.net ISSN: 2395-5252

defined to extract multiple features and assign the V. ARCHITECTURE OF PROPOSED

correct sentiment score to those features. MODEL
The Bayesian Classification is a supervised Architecture design gives the real world view of
statistical method for classification and contains the system.
practical learning algorithms. The posterior Data Flow Diagram represents the flow of
probability of a class can be computed using Naive information or data in the system. The square boxes
Bayes model. This model works is suitable for a represent the entities, ovals represent the process
large data set. The use of the Bayes Theorem is to and named arrows represent the direction of flow
presume the chance of the inclined feature set of information. Fig-1 shows the data flow diagram
matches to specified label. Bayes theorem provides of web scrapping Sentiment analysis system. It has
a way of calculating the posterior probability, two entities, WebCrawler and Classifier and
P(L|F), from P(L), P(F), and P(L|F). Naive Bayes threeprocesses namely Web Scraping,
classifier assumes that the effect of the value of a Preprocessing and Product Review.
predictor (F) on a given class (L) is independent of
the values of other predictors. [8]

Fig – 1 Data Flow Diagram

Use Case Diagram is used show the interaction of user and system. The use cases and actors in use case
diagram describe what the system does and how the actors use it, but not how the system operates internally.
Fig-2 show actors as user and system and oval represents process.

Fig-2 Use Case Diagram

DOI: 10.35629/5252-040513071311 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1309
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 5 May 2022, pp: 1307-1311 www.ijaem.net ISSN: 2395-5252

VI. RESULTS

Output of the proposed model

DOI: 10.35629/5252-040513071311 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1310
 International Journal of Advances in Engineering and Management (IJAEM)
Volume 4, Issue 5 May 2022, pp: 1307-1311 www.ijaem.net ISSN: 2395-5252

VII. ADVANTAGES indebted to our guide, Er. Akshay Tiwari and Er.
ANDDISADVANTAGES Ankit Khare, Department of Information
 By using sentiment analysis, you gauge how Technology, for innumerable acts of timely advice,
customers feel about different areas of your encouragement and we sincerely express our
product without having to read thousands of gratitude. We also thank her for her constant
customer comments directly. encouragement and support extended throughout.
 If you've thousands of feedbacks per month, Finally, yet importantly, we express our heartfelt
one person can’t read all of these responses. thanks to our family & friends for their wishes and
By using sentiment analysis and automating encouragement throughout our work.
this process you'll easily drill down into
different customer segments of your business REFERENCES
and procure a much better understanding of [1]. Raheesa Safrin, K. R. Sharmila, T. S. Shri
sentiment in these segments. Subangi, E.A. Vimal “Sentimental Analysis
 The structure of websites change frequently - on online product review”, 2017
Scraped data is arranged according to the [2]. Aarti A. Patil, “Sentimental Analysis of
structure of the website. Sometimes you Product review,” 2014
revisit the site and will find the ultimate layout [3]. “Comparative Analysis of Sentimental
changed. Orientation using SVM and Naïve Bayes
 The downsides of Streamlit are its lack of Techniques” by ShwetaRana, 2016
design flexibility and control over your [4]. Sujuta L. Sonavane, “Extracting a sentiment
application layout. from Review: A Lexicon Based Approach,”
 If your application and/or dataset is large, 2017.
you’ll likely run into speed issues due to the [5]. Sasikala , L.Mary Immaculate Sheela,
application flow - with the entire source code Research Scholar, Department of Computer
being re-run on every new change or science, Mother Teresa Women’s
interaction. University, Kodaikanal, India. “Sentiment
Analysis and Prediction of Online Reviews
with Empty Ratings” ISSN 0973-4562
VIII. CONCLUSION Volume 13, Number 14 (2018) pp. 11525-
Sentimental analysis and opinion mining
11531
is a field of study that analyzes people’s sentiment,
[6]. Python (programming language) -
attitude’s or emotions towards certain entities.This
https://en.wikipedia.org/wiki/Python_(progr
project on sentimental analysis on product review
amming_language)
using web scrapping tackles a fundamental
[7]. Inferring networks of substitutable and
problem of sentiment analysis, sentiment polarity
complementary products J. McAuley, R.
categorization. Online product reviews from E-
Pandey, J. Leskovec Knowledge Discovery
commerce website flip cart are selected as the input
and Data Mining, 2015
to this project. This will enable the system to
[8]. Evandro Costa, Rafael Ferreira, Patrick
analyze and then generate results (for reviews)
Brito, IgIbertBittencourt, OlavoHolanda,
which have been sorted and display in the form of
Aydano Machado, TarsisMarinho “A
charts. Possible future work is to improve the
framework for building web mining
efficiency.
applications in the world of blogs: A case
study in product sentiment analysis”, Expert
IX. ACKNOWLEDGEMENT Systems with Applications (39),pp.4813-
It is with great satisfaction and euphoria 4834,2012.
that we are submitting the paper on “Sentiment
Analysis on Product Reviews”. We are profoundly

DOI: 10.35629/5252-040513071311 Impact Factor value 7.429 | ISO 9001: 2008 Certified Journal Page 1311