Music Recommendation Using Conversational Recommender System with Explanation Facility

Abstract

The majority of music recommendation systems use user data such as ratings, likes, feedback, music played as the user model. This is certainly not suitable for new users and users who are new to music. In addition, most of the research related to music recommendation systems focuses on accuracy and pays less attention to user experience. This can be overcome by using an explanation facility which is the reason the system provides recommendations. Explanation facility is able to improve user experience in several factors such as Transparency, Scrutability, Trustworthiness, Effectiveness, Persuasiveness, Efficiency, and Satisfaction. Therefore, in this study, we develop a conversational recommender system with an explanation facility in the music domain. This system works by conducting a conversation between the system and the user, by a chatbot. This system is a knowledge-based recommender system and uses an ontology that serves as a knowledge base for recommending music. We evaluate the system based on two parameters, such as recommendation accuracy and also the influence of explanation for user satisfaction. Tests show that a system with an explanation facility increases user experience more than a system without an explanation facility. In addition, system testing also shows high accuracy (90.48 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> ) and successfully meets user needs.