Hybrid Music Recommendation System Based on Temporal Effects

Abstract

Use of music recommendation system is fully grown because of large number of online music websites. The most challenging gap we found is that there is an utmost need to consider more contextual information like weekdays, session of day, time, and frequency of listening songs. In this paper, we propose a hybrid music recommendation system which is a combination of two approaches. In the first approach we propose to use cosine similarity measure for ranking of music. The second approach considers the graph-based approach. In the graph-based approach we propose using particle swarm optimization with differential evolution to get optimized ranking of music. We recommend top-n songs by combination of these two approaches. Standard Last.fm dataset is considered for experimental purpose. Data pre-processing operation is performed on dataset to remove the noisy and inconsistent data. Comparison of our proposed model with the state-of-the-art model shows the effectiveness in the form of recall rate.