Combining User-Genre Preference Information with Neural Collaborative Filtering

In this paper, the user-genre preference information is introduced into collaborative filtering (CF) to deal with the sparsity which CF suffers from. This work classifies items by genre information, it applies neural collaborative filtering model in each category and combines all models together to get a final prediction. These experiments in the work are conducted on well-known dataset in which hit ratio and Normalized Discounted Cumulative Gain (NDCG) are introduced into the evaluation. It indicates that our model has fast performance growth and good results.