Enhancement of AI music recommendation system to promote diversity and mitigate the cold start problem through hybrid model of social ACT-R and Gru4Rec with collaborative filtering

Abstract

This thesis delves into the existing algorithms used in recommendation system, particularly Music Recommendation System which have been widely in use to suggest music to the users. To address the diversity enhancement in music consumption, this paper represents the analysis of algorithms widely used by mainstream music streaming services and supporting social ACT-R and GRU4Rec model in collaboration with collaborative filtering. The purpose is to root for the domain of cognitive theory to cover the wide range of music genre to enhance the diversity and inclusivity to the Music RS (Recommendation System). It also aims to mitigate the cold start problem for the new user. In the context of boosting diversity and bridging the gap of cold start problem, this paper adopted the research methodology of theoretical analysis based on literature review done on potential models and framework. To illustrate the findings a logical explanation has been conducted to analyse the dataset and algorithm which illustrates the case studies and emphasise on the hybrid model of Social ACT-R and GRU4Rec with collaborative filtering.