An Efficient Neighborhood-based Interaction Model for Recommendation on Heterogeneous Graph | AISC
Subscribers:
22,300
Published on ● Video Link: https://www.youtube.com/watch?v=Vkdz-wn6vFw
For slides and more information on the paper, visit https://ai.science/e/graph-nn-rec-sys-an-efficient-neighborhood-based-interaction-model-for-recommendation-on-heterogeneous-graph--d1I1BXudpzaEGHGwLXEI
Speaker: Jiarui (Jerry) Jin; Host: Omar Nada; Discussion Facilitator: Nabila Abraham
Motivation:
Obtaining state-of-art accuracies on Recommender Systems using Graph Networks by avoiding the downfalls of early aggregation is brand novel work that can be used in lot more Recommender Systems problems and datasets. Combining different types of nodes (Actors, Movies, Director, etc.) and then applying a specific random walk to find the best movies for a given user by avoiding the early aggergation.