LightGCN: linear embedding propagation for collaborative filtering

What changed

Simplified graph collaborative filtering to normalized linear neighborhood embedding propagation and weighted combination across depths, omitting intermediate feature transforms and nonlinear activations.

Scientific scope

Simplified graph collaborative filtering to normalized linear neighborhood embedding propagation and weighted combination across depths, omitting intermediate feature transforms and nonlinear activations. Distinct from SGC's attributed-node linear classifier; here learned ID embeddings and user–item interaction graphs serve collaborative filtering. Source ablations do not prove nonlinear features useless in all recommender settings. Original template Conference17/placeholderDOI is not actual2017publication.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Distinct from SGC's attributed-node linear classifier; here learned ID embeddings and user–item interaction graphs serve collaborative filtering. Source ablations do not prove nonlinear features useless in all recommender settings. Original template Conference17/placeholderDOI is not actual2017publication.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction