Sentimeter: A Sentiment Analysis Platform For

Product Reviews
Ms. Mansi Aggarwal Prakhar Pratap Singh
BTech CSE AIML BTech CSE AIML
ABES Engineering College ABES Engineering College Prachi Singh
Ghaziabad, India Ghaziabad, India BTech CSE AIML
ABES Engineering College
Ghaziabad, India
Mudassir Imam Pradeep Kumar Yadav
BTech CSE AIML BTech CSE AIML
ABES Engineering College ABES Engineering College
Ghaziabad, India Ghaziabad, India

Abstract— Sentiment analysis has emerged as a critical tool II. METHODS AND MATERIAL
in understanding user perceptions and product feedback.
Sentimeter is a sentiment analysis platform designed for A. Tools
Flipkart product reviews. It provides users with sentiment
scores, a pie chart showing sentiment distribution (positive, 1. Frontend:
negative, neutral), geographical sentiment analysis, a word ReactJS is utilized for building responsive and dynamic
cloud of keywords, and an AI-generated summary. user interfaces, while ChartJS is employed to visualize data
Additionally, it enables comparative analysis of two products. effectively through pie charts and graphs. Additionally,
The project leverages cutting-edge tools and technologies, Framer Motion is used to add smooth animations, enhancing
including ReactJS for the frontend, ExpressJS for the backend,
the overall user experience.
and Python with Natural Language Toolkit (NLTK) for data
analysis. This paper outlines the project’s architecture, 2. Backend:
methodology, tools, and potential applications for consumers
and product sellers. ExpressJS functions as the backend server, efficiently
handling requests and routing data.
Keywords—Sentiment Analysis, Natural Language
3. Data Analysis:
Processing, E-commerce, ReactJS, ExpressJS, NLTK
Python serves as the core language for data analysis and
I. INTRODUCTION extraction, with the Requests module facilitating HTTP
requests to fetch data. Beautiful Soup is employed to extract
In an era where e-commerce dominates the market, product reviews, titles, and images from Flipkart. For natural
customer reviews play a pivotal role in influencing purchase language processing, NLTK provides tools such as the
decisions. However, analyzing thousands of reviews VADER Lexicon, which calculates polarity scores to
manually is impractical. Sentimeter addresses this challenge determine sentiment intensity, and Stopwords and
by automating review analysis and providing actionable Tokenizers, which filter common words to identify and
insights for both consumers and sellers. process frequent keywords. Additionally, the Groq API
Sentimeter was developed to address this issue by generates AI-driven summaries of reviews.
automating sentiment analysis for Flipkart product reviews.
By providing users with sentiment scores, distribution charts,
and concise summaries, the platform simplifies the process of
understanding user feedback. The integration of advanced B. Methodology
visualization tools further enhances user experience by The workflow of Sentimeter is organized into three main
presenting data in an intuitive and actionable format. stages: data extraction, analysis, and visualization. In the data
extraction phase, users input a Flipkart product review page
The platform’s capability to perform geographical sentiment URL on the frontend. This URL is sent to the backend server,
analysis and word cloud generation empowers users and built with ExpressJS, via a POST request. The backend then
sellers alike. While consumers can make informed purchase calls a Python function that uses Beautiful Soup to scrape
decisions, sellers gain insights into their products' reviews, locations, and product metadata.
performance across different regions. Additionally, the
comparative analysis feature enables quick evaluation of During the data analysis stage, the extracted reviews are
competing products. processed using NLTK’s VADER Lexicon, which assigns
polarity scores to individual reviews and calculates an
This paper explores the architecture, tools, and average sentiment score. Reviews are categorized into
methodologies that make Sentimeter a comprehensive positive, negative, and neutral sentiments, creating a
sentiment analysis solution. By leveraging technologies like distribution. Additionally, location-based sentiment scores
ReactJS, ExpressJS, and NLTK, Sentimeter sets a benchmark are generated for geographical insights, and frequent
for scalable and efficient sentiment analysis in the e- keywords are identified through tokenization and stopword
commerce sector. removal. The Groq API further synthesizes a concise
summary of all reviews for a comprehensive understanding.
 In the visualization stage, the results are dynamically
displayed to users. Sentiment scores, distributions, and
geographical data are visualized using ChartJS, while word
clouds are generated to emphasize frequently mentioned
keywords. Comparative analysis of the results, including a
summary of the better product, is also presented to assist
users in making informed decisions.

III. RESULTS AND DISCUSSION

The architecture of Sentimeter adopts a modular design to
ensure functionality and scalability. The frontend, built with
ReactJS, captures user input and dynamically displays the
results, offering an intuitive user interface. The backend,
implemented using ExpressJS, handles API requests and
responses, acting as a bridge between the frontend and the
data processing modules. A Python model executes the core
tasks of data extraction and analysis, utilizing libraries like
NLTK to process and interpret the data. This interaction IV. CONCLUSION
between components ensures a seamless and efficient user Sentimeter demonstrates the potential of combining web
experience. technologies and natural language processing to deliver
Sentimeter delivers various visualization outcomes to actionable insights from e-commerce reviews. By automating
present data-driven insights effectively. Sentiment the analysis process, the platform saves time and effort for
distribution is showcased through pie charts, illustrating the both consumers and sellers. Future work may include
percentage of positive, negative, and neutral reviews for a integrating multilingual support, expanding the geographical
product. Geographical sentiment analysis further enriches the analysis to more regions, and incorporating additional e-
insights by highlighting trends based on user locations, commerce platforms. With its robust methodology and user-
enabling a deeper understanding of regional perspectives. centric design, Sentimeter is a step forward in democratizing
sentiment analysis.
To enhance comprehensibility, a word cloud is generated to
emphasize frequently mentioned keywords from the reviews.
Additionally, Sentimeter provides comparative analysis by REFERENCES
aggregating cumulative sentiment scores, helping users [1] Sentiment Analysis of Product Reviews by Cane W. K. Leung, and
identify the better product based on user feedback. These Stephen C. F. Chan, The Hong Kong Polytechnic University, Hong
Kong SAR
visualizations and analyses empower users to make informed
[2] A Study of Sentiment Analysis: Concepts, Techniques, and
decisions with clarity and confidence. Challenges by Ameen Abdullah Qaid Aqlan
The interaction flow is as follows: [3] Sentiment analysis methods, applications, and challenges: A
systematic literature review by Yanying Mao
[4] Sentiment Analysis Using the Vader Model
[5] Sentiment Analysis Optimization Using Vader Lexicon on Machine
Learning Approach by Dhendra Mutho
[6] NLTK: the Natural Language Toolkit by Edward Loper and Stever
Bird.
[7] Analysis of customer reviews with an improved VADER lexicon
classifier by Kousik Barik & Sanjay Mishra