A Survey on Accuracy-oriented Neural Recommendation: From Collaborative Filtering to Information-rich Recommendation

Influenced by the great success of deep learning in computer vision and language understanding, research recommendation has shifted to inventing new recommender models based on neural networks. In recent years, we have witnessed significant progress developing models, which generalize surpass traditional owing strong representation power this survey paper, conduct a systematic review aiming summarize field facilitate researchers practitioners working systems. Specifically, data usage during modeling, divide work into collaborative filtering information-rich recommendation: 1) filtering, leverages key source user-item interaction data; 2) content enriched recommendation, additionally utilizes side information associated with users items, like user profile item knowledge graph; 3) temporal/sequential accounts for contextual an interaction, such as time, location, past interactions. After reviewing representative each type, finally discuss some promising directions field.