Generating the Structured Answer for Feature-based Sentiment Questions from Customer Reviews

Abstract

With the rapid growth the rich opinionated information on various features of product, people are more interested in opinion questions that can immediately reflect others’ opinions. Opinion question answering is an important research area in opinion mining and Natural Language Processing. Question analysis and answer generating are also the important keys for this system. A new approach, pattern knowledge in linguistic with the POS tagging are developed for feature based sentiment questions analysis on the products. This paper also introduces a Wordnet based methodology to retrieve the review sentences from the customer reviews. Finally, the system aggregates the quality review sentences by using polarity grouping and a simple ranking approach to get the complete set of answer for sentiment questions