An Approach to Incorporate Textual Reviews in Book Recommender System

Abstract

Recommender systems are widely accepted by internet users to find suitable products in various domains. In the academic domain, the book recommender system provides personalized books to the learner to retain his interest in the learning environment. It also helps in reducing cognitive stress faced by learners due to non-personalized recommendations and information overload. The existing recommender systems recommend books to the learner based on the learner’s preferences yet the learner struggles in retrieving tailored recommendations. This leads to an increase in learner’s stress and anxiety. To relieve stress and anxiety, appropriate recommendations can be provided to the learner by considering the textual reviews of books given by peer learners. Sentiments of the textual reviews can be used to complement the explicit ratings of books to achieve higher recommendation reliability. So the objective of this paper is to propose and elaborate the framework for a book recommender system that integrates sentiment analysis with collaborative filtering to improve the predicted ratings. The proposed recommendation system is based on KNN and lexicon-based sentiment analysis. The results of the evaluation and testing of the proposed framework show that there is a significant improvement in the performance of the recommender system.