Simplices-based higher-order enhancement graph neural network for multi-behavior recommendation

Qingbo Hao, Chundong Wang, Yingyuan Xiao, Hao Lin

Abstract: Highlights•We propose a novel higher-order enhancement GNN for multi-behavior recommendations.•We explore the impact of higher-order relations on multi-behavior recommendations.•We define the implicit relationship calculation method to alleviate data sparsity.•We introduce simplex complexes to capture higher-order relations.