PriceWatch360: Product Price Comparison & Sentiment Analysis

In the current scenario, it is difficult to figure out whether the price for any product displayed on the website is the best possible price or not, and surfing on different sites for the same product to get a reasonable price is a tedious job. To overcome this problem, we put forward this system. This research study presents a system developed by the authors, which integrates web scraping for price comparison and machine learning for sentiment analysis, aimed at enhancing consumer decision-making. Utilizing Python and libraries such as Beautiful Soup and Requests, product details are extracted from Amazon and Flipkart. Selenium handles dynamic content, ensuring comprehensive data retrieval for accurate price comparison. Discount percentages are calculated to aid in price assessment, while sentiment analysis is performed using a Support Vector Machine (SVM) model will be trained upon the Amazon Reviews dataset from Kaggle. Data pre-processing steps include cleaning, tokenization, and transforming text into numerical features using TF-IDF. The SVM model is then trained on labelled review data, enabling it to classify sentiments as positive, negative, or neutral. To ensure proper sentiment classification, the efficiency of the model is evaluated using parameters including accuracy, precision, recall, and F1-score. Cross-validation techniques further enhance the model's robustness. An average sentiment score is also calculated to provide an overall sentiment perspective. The integration of web scraping and sentiment analysis functionalities is achieved through a Flask-based web application. This intuitive user interface allows users to search for products, compare prices, and view sentiment analysis results. Upon entering a product name or keyword, the application scrapes data from Amazon and Flipkart, retrieves product details and prices, and performs sentiment analysis on user reviews, presenting the results in an accessible format.