Hyperbolic spatial-temporal network for session-based recommendation

Jiaxin Zhang

doi:10.1016/j.asoc.2025.113996

Hyperbolic spatial-temporal network for session-based recommendation

2025-09-29

0

PRE

AI

DOI：10.1016/j.asoc.2025.113996

Original

Original request for help

Save

Abstract

En 中文

• We introduce a hyperbolic hypergraph neural network as the base encoder to alleviate the skewed long-tail distribution and to enhance the quality of item representations in hyperbolic space. • We designed a spatial-temporal mining module that can simultaneously learn spatial features and temporal features within session data to generate the positive samples for contrastive learning, enhancing the ability to learn and represent these two features in the model. • Extensive experiments on various datasets (Tmall, Diginetica, Nowplaying) demonstrate its superiority compared to existing methods and provide insights for future research in session-based recommendation.

Hyperbolic spatial-temporal network for session-based recommendation

Abstract

Journal

Organization

Cited Papers