📄Final Year MAJOR Project Documentation

E-COMMERCE PRODUCT REVIEWS’ SENTIMENT ANALYSIS

• 🗓️ Project Duration: 20th September 2024 – 30th June 2025
• 👨‍🏫 Mentor: Dr. Mrinmoy Sen (Assosiate Professor at HALDIA INSTITUTE OF TECHNOLOGY )
• 👥 Team Size: 5
• 🔗 Live link: Deploy Link
• 🔗 Project Repository: GitHub - Amazon Reviews Sentiment Analysis

🔖 About the Project

This final year major project focuses on analyzing customer feedback through Sentiment Analysis of e-commerce product reviews, aiming to improve user experience and assist businesses in making data-driven decisions. By leveraging Machine Learning (ML) and Natural Language Processing (NLP) techniques, the project extracts actionable insights from textual data, highlighting customer satisfaction trends and areas of improvement.

The project includes the development of a Streamlit-based web application, which enables users to:

• Analyze single-line reviews for quick sentiment classification

• Upload and process bulk reviews via CSV files

• 

• 

• 

• 

🎯 Objectives

• To conduct sentiment analysis on e-commerce product reviews to understand customer perception

• To enhance marketing strategies by analyzing trends in customer feedback

• To develop a user-friendly tool for visualizing customer sentiments and insights

• To research and address challenges like handling unstructured data, emotional nuances, and dataset biases

🛠️ Key Skills & Technologies

• Python

• Pandas

• NLP (Natural Language Processing)

• NLTK (Natural Language Toolkit)

• Logistic Regression

• Machine Learning Algorithms

• Streamlit (Web Application Framework)

• Web Scraping (for Amazon Reviews)

🏗️ Project Architecture & Workflow

1. Data Collection

   • Manual review entries

   • CSV file uploads (bulk data)

   • Scraped data from Amazon product pages using BeautifulSoup or similar tools

2. Preprocessing

   • Tokenization

   • Stop-word removal

   • Lemmatization / Stemming

   • Handling emojis and special characters

   • Removing noise from unstructured data

3. Feature Extraction

   • Bag-of-Words

   • TF-IDF Vectorization

4. Model Training & Evaluation

   • Applied Logistic Regression for classification

   • Experimented with various ML classifiers (Naive Bayes, SVM)

   • Explored deep learning models (RNN/LSTM) as potential future enhancements

   • Performance evaluated using accuracy, precision, recall, and F1-score

5. Sentiment Analysis Categories

   • Positive

   • Negative

   • Neutral

6. Visualization & Dashboard

   • Developed with Streamlit

   • Displays charts, graphs, and summary insights for better understanding

💡 Key Contributions & Outcomes

• Conducted extensive research on state-of-the-art methods, including rule-based approaches, traditional ML models, and deep learning

• Addressed challenges such as dataset biases and emotional subtleties in sentiment classification

• Delivered an interactive Streamlit app for end-users to analyze and visualize sentiment trends

• Provided actionable insights to help businesses improve product quality and enhance user satisfaction

🚀 Future Scope & Enhancements

• Integrating real-time sentiment analysis for live product reviews

• Adding multi-language support for global customer feedback

• Implementing deep learning models (LSTM, BERT) for higher accuracy and contextual understanding

• Expanding the platform to include voice-based review analysis

• Creating a mobile app version for on-the-go insights

👨‍💻 Team

• @SayanRony
• @srinjoy29
• @Sayanijana23
• @pallav34
• @Nilambargiri