The role of sentiment analysis in a recommender system: a systematic survey. (19th August 2022)

• Record Type: Journal Article
• Authors: Patel, Jitali; Chhinkaniwala, Hitesh
• Is Part Of: International journal of Web engineering and technology. Volume 17:Number 1(2022)
• Page Start: 29
• Page End: 62
• Publication Date: 2022-08-19
• Subjects: recommender system -- sentiment analysis -- user-generated text -- collaborative filtering -- sentiment-aware recommender system; World Wide Web -- Periodicals; Web site development -- Periodicals; Application software -- Development -- Periodicals; 006.7
• Languages: English
• ISSNs: 1476-1289

Abstract

Currently, fields like e-commerce, education, social media, tourism, and the entertainment industry rely on recommender systems to provide personalised services to their clients. The most common and widely accepted technique - collaborative filtering, creates recommendations by examining the users' past rating patterns. Collaborative filtering assumes that a users' past rating data accurately reflects their actual preferences. However, different study found that the ratings may not accurately reflect user preferences in the real-world circumstances. Therefore, to deal with this problem, sentiment analysis of user-generated text is started to be used. It helps to improve the performance of recommender systems, as it provides more specific and trustworthy user preferences than ratings. A sentiment-aware recommender system captures sentiment from the user-generated content and provides most suited personalised services to the user. We have classified sentiment enhanced recommender systems according to the level of sentiment analysis and presented technical aspects such as datasets, methodologies and results.