Deep Content-User Embedding Model for Music Recommendation
This work addresses the cold-start problem for music recommendation systems, offering an incremental improvement by integrating collaborative and content-based filtering in a joint training framework.
The authors tackled the cold-start problem in music recommendation by proposing an end-to-end deep content-user embedding model that jointly trains on user-item interactions and music audio content, achieving significant performance improvements over previous methods.
Recently deep learning based recommendation systems have been actively explored to solve the cold-start problem using a hybrid approach. However, the majority of previous studies proposed a hybrid model where collaborative filtering and content-based filtering modules are independently trained. The end-to-end approach that takes different modality data as input and jointly trains the model can provide better optimization but it has not been fully explored yet. In this work, we propose deep content-user embedding model, a simple and intuitive architecture that combines the user-item interaction and music audio content. We evaluate the model on music recommendation and music auto-tagging tasks. The results show that the proposed model significantly outperforms the previous work. We also discuss various directions to improve the proposed model further.