Discovering a Wide Range of Features For Sentiment Quantification Using Naive Bayes Algorithm

Abstract

Today, there are several product reviews accessible via the web. Consumer reviews provide a large amount of informative details that is useful to both consumers and businesses. However, product reviews are often unorganized, causing problems with collection of data and management of knowledge. The aim of this document is to present a grading system for product features that spontaneously identifies key features of products based on online customer feedback, with the goal of enhancing the usability of the various reviews. The most significant product aspect is characterized by two findings. First, a decent number of customers typically remark on the most critical things, and second, customer views on the important considerations have a substantial impact based on their general opinions of the product in terms of aspects rating. Given the customer feedback, we use a shallow dependency parser to classify reviews of product aspects and a sentiment classifier to assess consumer opinions on these aspects ranking. Then, by simultaneously considering the frequency of the aspect as well as the effect of customer opinions provided to each element of their entire product review viewpoints, we create a probability-based aspect ranking algorithm to infer the value of aspects.