Neural Collaborative Filtering for POI Recommendation

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "Neural Collaborative Filtering for POI Recommendation"

by

YAP Alistair Yun Hee

Abstract:

Traditional recommender systems utilize collaborative filtering to predict 
each user's preference for items they have not interacted with. However, 
collaborative filtering alone is incapable of explicitly incorporating 
external factors besides preference. In particular, studies into Point of 
Interest recommendation for Location Based Social Networks have identified 
3 other influential factors: social influence, temporal trends and 
spatiotemporal constraints.

In this work, we investigate the significance of each factor when utilized 
in several variations of neural collaborative filtering architecture. We 
empirically find incorporating spatial and temporal factors to provide 
significant improvement in recommendation quality, while all attempts to 
utilize social influence failed to yield any improvements. Furthermore, we 
empirically find that simple neural collaborative filtering models 
outperform deeper, more complex models.


Date            : 2 May 2019 (Thursday)

Time            : 17:00 - 18:00

Venue           : Room 4621 (near lifts 31/32), HKUST

Advisor         : Prof. CHAN Shueng-Han Gary

2nd Reader      : Dr. HUI Pan