Neural Graph Collaborative Filtering

TLDR

Learning vector representations (embeddings) of users and items is central to modern recommender systems, evolving from matrix factorization to deep learning, and typically mapping from pre‑existing features such as IDs and attributes. The authors argue that existing methods fail to encode collaborative signals from user‑item interactions into embeddings. Consequently, the resulting embeddings may inadequately capture collaborative filtering effects.

Abstract

Learning vector representations (aka. embeddings) of users and items lies at the core of modern recommender systems. Ranging from early matrix factorization to recently emerged deep learning based methods, existing efforts typically obtain a user's (or an item's) embedding by mapping from pre-existing features that describe the user (or the item), such as ID and attributes. We argue that an inherent drawback of such methods is that, the collaborative signal, which is latent in user-item interactions, is not encoded in the embedding process. As such, the resultant embeddings may not be sufficient to capture the collaborative filtering effect.