POLIBITS, Vol. 57, pp. 39-43, 2018.
Abstract: review mining, dataset, sentiment analysis, features, parts of speech tagging, opinion word, dependency parsing
Keywords: Reviews are a valuable resource. Conclusions drawn on analysis of reviews are of great help in improving the product, as far as the manufacturer is concerned, or with predicting sales figures, as far as the retailer is involved. However, employing human labor to go through all the reviews manually would be a time consuming and expensive process. This paper outlines a novel technique to extract features from a product’s reviews along with the corresponding sentiment expressed, using POS tagging and Dependency Parsing in conjunction. The use of these of these allows both the context and the parts of speech of a word to be employed in feature and corresponding opinion word detection. The opinion word is given a sentiment polarity determined from a training set of positive and negative reviews. The method described in this paper is for large data sets, and requires no domain specific data for feature extraction.
PDF: Simple and Effective Feature Based Sentiment Analysis on Product Reviews using Domain Specific Sentiment Scores
PDF: Simple and Effective Feature Based Sentiment Analysis on Product Reviews using Domain Specific Sentiment Scores
https://doi.org/10.17562/PB-57-3
Table of contents of POLIBITS 57