Feature based mobile phone rating using sentiment analysis and machine learning approaches

Abstract

This project proposes a model of sentiment analysis of different features of different company’s mobile sets and rating them overall. Customers before buying a phone check reviews to get a better understanding of the device and this project derives an optimum solution for this. In this model, every feature of a mobile phone is rated based on public opinion and an overall rating for every type. Amazon is one of the largest internet retailer, which makes way for most public reviews on their products and so we collected data for sentiment analysis from amazon. We pre-processed the gathered data to a supervised form and chose the most common features from train data. In our model, Naïve Bayes, Support Vector Machine, Logistic Regression, Stochastic Gradient Descent and Random Forest algorithms were used to compare performance. These classification algorithms were trained with the training data and tested with test dataset to determine the accuracy of the classifiers. Our model provides an average polarity of each features and an average polarity of the mobile phone which will give a rating of the device, thus assisting the customers to choose the best according to their desire. This project can work as an assistant for the customers to determine their device following the opinion of the other users of the device.