Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 11

INTERNATIONAL CONFERENCE ON

INNOVATION IN ENGINEERING &


SUSTAINABILITY THROUGH CREATIVITY AND INNOVATION
MANAGEMENT (ICIEM’23)
PAPER TITLE:
Sentiment analysis with the Amazon product reviews using Python
Nikhil Singh
Student, Noida Institute of Engineering and Technology, nikhilcse053@gmail.com
Tamanna
Research scholar, Noida Institute of Engineering and Technology, tamannaveer22@gmail.com
Rahul Srivastava
Student, Noida Institute of Engineering and Technology, officialrahulsrv@gmail.com
Nilay Pandey
Student, Noida Institute of Engineering and Technology, np.nilay21@gmail.com
Sayeed Ahmad khan
Student, Noida Institute of Engineering and Technology, sayeedahmadkhan663@gmail.com

BIRLA INSTITUTE OF TECHNOLOGY, MESRA (RANCHI), NOIDA CAMPUS, INDIA: A-7, Sector 1, Noida – 201301, India | www.bitmesra.ac.in/bitnoida
Abstract
CONTENT
Introduction
Methodology
Results
Conclusion
Future Work
ABSTRACT
• Sentiment analysis is the process of analyzing opinions, experiences, and emotions from reviews of any
product or service. One of the important issues with sentiment analysis of product reviews is that they
produce a summary of opinions based on a pre-built English dictionary. As a result, reviews expressed in
regional languages have not been identified or considered as neutral opinions. This reduces the accuracy
of the sentiments built based on reviews. In this study, the local language terms/words used for Amazon's
specific product reviews have been incorporated into the Python nltk toolkit. The sentiment analysis
based on the modified approach is categorized in terms of positive, negative, and neutral and is
compared with the previous model. The sentiment analysis is carried out by determining the polarity of
the text, obtaining its precise context, and correct classification. The analyzed data helps customers in
decision-making for a specific product and manufacturer in robust design.
INTRODUCTION
1. Generally, e-commerce platforms ask users to review their products. The reviews typically appear as users' viewpoints in
terms of positive, negative, or neutral feedback.
2. These viewpoints assist other users/customers form opinions and make decisions regarding a specific product, whether for
businesses, governments, or individuals.
3. However, manual analysis of reviews and categorizing them as positive, negative, or neutral is difficult due to the vast
amount of data. Thus, an automated method is essential to address this issue.
4. Sentiment analysis is usually employed to determine the polarity of the text, obtain its precise context, and provide accurate
classification.
Methodology
• Uploaded review data, in CSV format acquired from the Amazon e-
commerce website.
• Uploaded lexicon from newly created lexicon file and updated the
current NLTK(Natural Language ToolKit) Library.
• Computed polarity of sentiments using NLTK with new data inside.
• Result Interpreted that it reduced Indian language review into
polarity instead of marking it neutral.
Results
Figure 1. Before updating lexicon VADER Figure 2. After updating lexicon VADER
RESULTS
• As observed the fig3 graph shows 0.0 neutral polarity
• Fig4 graph showing negative polarity.
• This shows that it interpreted a neutral review between Sept-2021 to Dec-2021 into a negative
review.
Conclusion
• This paper contributes to improving the opinion mining model that can
identify reviewed sentences from an extensive collection of reviews and
calculate the review's polarity with good accuracy and better comparison to
the existing models and algorithms.
• As it consists of locally used words frequently used by reviewers, its
accuracy is more than before as we show in the results.
FUTURE WORK
1. In the future, it could contribute to predicting the performance of products in the market, and companies
can use it to predict how their products will fare in the market.
2. Our study introduced a more precise lexicon VADER in natural language libraries, which can provide a
proper glimpse of products, neutralize neutral polarity scores, and derive sentiments from them.
3. Also in the future lexicon VADER understand emoji reactions to get even more precise sentiments from
product reviews.
THANK YOU

You might also like