Aspect Extraction and Sentiment Analysis on E-commerce reviews using Deep Learning

For the better understanding of user insights, sentimental analysis has been the most important tool. Our project is based on the sentimental analysis of user reviews generated from various e-commerce websites with the help of Large Language Models(LLMs) which has been implemented with the means of Ollama framework. With the development of an advanced system that helps in extracting user reviews followed by analysis of patterns on text, overall sentiment determination and finally performing apsect based analysis on the same, this model helps us to categorize customer comments into various aspect that is self generated by the model. The obtained insights is integrated in a user interface which is built with Streamlit, which provides us with extraction of text, generation of structured summary and general sentiment computation. The observations are then stored in a NoSQL database, CouchDB, for easier and hassle free retrieval of data. Finally a chrome extension was created using FastApI, which pops up a notfication whenever an user searches for a product on the e-commerce website Flipkart, providing a detailed product summary of various aspects analyzed along with the final overall sentiment. Overall this study provides us with better understanding of the effective working of the LLM model in the field of sentimental analysis and can be further used to provide useful information on performance of the product to users and businesses.