Abstract
LightGCN including only the most essential component in GCN-neighborhood aggregation- for collaborative filtering.
LightGCN learns user and item embeddings by linearly propagating them on the uwer-item interaction graph, and uses the weighted sum of the embeddings learned at all layers as the final embedding.
Such simple, linear and neat model is much easier to implement and train, exhibiting substanctial imporovements over NGCF under exactly the same experimental setting.
Introduction
NGCF takes inspiration from the GCN, following the same propagation rule to refine embeddings; feature transformation, neighborhood aggregation, and nonlinear activation. But we argue tha its designs are rather heavy and burdensome. and they are not necessarily useful for the CF task.
To be specific, GCN is originally proposed for node classification on attributed graph, where each node has rich attributes as input features; whereas in user-item interaction graph for CF, each node (user or item) is only decribed by a one-hot ID, which has no concrete semantics besides being an identifier. In such a case, given the ID embedding as the input, performing multiple layers of nonlinear feature transformation - will bring no benefits, but negatively increase the difficulty for model training.
To validate our thought, we perform extensive ablation studies on NGCF. With rigorous controlled experiments, we draw the conclousion that the two operations inherited from GCN-feature transformation and nonlinear activation-has no contribution on NGCF's effectiveness. Even more surprsing, removing them leads to significant accuracy improvements.
Finally we represent a new model named LightGCN, including the most essential component of GCN - neighborhood aggregation - for collaborative filtering.
Specifically, after assiciating each user(item) with and ID embedding, we propagate the embeddings on the user-item interaction graph to refine them.
각 유저(Item)-ID embedding 한 후 user-item interaction grpah 에 임베딩 전파하여 구체화. Then combine the embeddings learned at diffenrent propagatioin layers with a weighted sum to obtain the final embedding for prediction. The whole model is simple and elegant, which not easier to train, but also achieves better empirical performance than NGCF and other state-of-the-art methods like Mult-VAE.
main contributions
- GCN 의 feature transformation, nonlinear activation 은 CF의 효과에 positive effect가 없음
- GCN의 핵심적 구성요소를 포함하면서도 simple한 모델인 LightGCN 개발
- 같은 환경에서 LightGCN과 NGCF를 비교했을 때 상당한 개선을 보임
