CombiGCN: An effective GCN model for Recommender System
Loc Tan Nguyen, Tin T. Tran
TL;DR
The paper addresses the limitation of single-graph propagation in GNN-based recommender systems by introducing CombiGCN, which propagates item embeddings on the user-item interaction graph via Light Graph Convolution and user embeddings on both the user-item and a newly constructed user-user weighted graph. A data-preprocessing step reduces noise and overfitting, and a simple weighted-sum fusion combines the two propagation streams for each layer. The approach leverages a Jaccard-inspired user similarity to build the user-user graph and uses a combined propagation with a final embedding ensemble optimized with Bayesian Personalized Ranking, showing superior performance on Ciao, Epinions, and Foursquare against strong baselines like NGCF, WiGCN, GCMC, and LightGCN. This results in improved embedding quality and recommendation accuracy, highlighting the practical value of multi-graph fusion and weighted collaborative signals in large-scale recommender systems.
Abstract
Graph Neural Networks (GNNs) have opened up a potential line of research for collaborative filtering (CF). The key power of GNNs is based on injecting collaborative signal into user and item embeddings which will contain information about user-item interactions after that. However, there are still some unsatisfactory points for a CF model that GNNs could have done better. The way in which the collaborative signal are extracted through an implicit feedback matrix that is essentially built on top of the message-passing architecture of GNNs, and it only helps to update the embedding based on the value of the items (or users) embeddings neighboring. By identifying the similarity weight of users through their interaction history, a key concept of CF, we endeavor to build a user-user weighted connection graph based on their similarity weight. In this study, we propose a recommendation framework, CombiGCN, in which item embeddings are only linearly propagated on the user-item interaction graph, while user embeddings are propagated simultaneously on both the user-user weighted connection graph and user-item interaction graph graphs with Light Graph Convolution (LGC) and combined in a simpler method by using the weighted sum of the embeddings for each layer. We also conducted experiments comparing CombiGCN with several state-of-the-art models on three real-world datasets.