A Hybrid Approach for Feature Extraction From Reviews to Perform Sentiment Analysis
Alok Kumar, Renu Jain · 2021
In this chapter, a hybrid approach to extract the important attributes called features of a product or a service or a professional from the textual reviews/ feedbacks has been proposed. The approach makes use of topic modeling concepts and the linguistic knowledge embedded in the text using Natural Language Processing tools. A system has been implemented and tested taking the feedbacks of two different domains: feedback of teachers and feedback of laptops. The system tries to extract all those features (single word and multiple words) for which users have expressed their opinion in the reviews. The syntactic category and the frequency contribute in deciding the importance of a feature and a numerical value between zero and one called weight is generated for each identified feature representing its significance. Results obtained from the proposed system are comparable if extraction from the text is done manually.