Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-view Hypergraph-based Contrastive Learning Model for Cold-Start Micro-video Recommendation

Sisuo Lyu, Xiuze Zhou, Xuming Hu

TL;DR

This paper tackles cold-start micro-video recommendation by integrating a multi-view multimodal feature extractor with a hypergraph-based embedding layer and cross-view contrastive objectives. It models higher-order user–item interactions via a hypergraph and aligns multimodal representations through cross-modal and graph-hypergraph contrastive losses, trained alongside a BPR objective. Experiments on MicroLens-50K and MicroLens-100K show that MHCR consistently outperforms baselines, with notable gains in Recall@10 and NDCG@10 and substantial improvements under cold-start conditions. The results demonstrate that combining hypergraph signal propagation with self-supervision on multimodal data yields robust recommendations in sparse, cold-start environments.

Abstract

With the widespread use of mobile devices and the rapid growth of micro-video platforms such as TikTok and Kwai, the demand for personalized micro-video recommendation systems has significantly increased. Micro-videos typically contain diverse information, such as textual metadata, visual cues (e.g., cover images), and dynamic video content, significantly affecting user interaction and engagement patterns. However, most existing approaches often suffer from the problem of over-smoothing, which limits their ability to capture comprehensive interaction information effectively. Additionally, cold-start scenarios present ongoing challenges due to sparse interaction data and the underutilization of available interaction signals. To address these issues, we propose a Multi-view Hypergraph-based Contrastive learning model for cold-start micro-video Recommendation (MHCR). MHCR introduces a multi-view multimodal feature extraction layer to capture interaction signals from various perspectives and incorporates multi-view self-supervised learning tasks to provide additional supervisory signals. Through extensive experiments on two real-world datasets, we show that MHCR significantly outperforms existing video recommendation models and effectively mitigates cold-start challenges. Our code is available at https://github.com/sisuolv/MHCR.

Multi-view Hypergraph-based Contrastive Learning Model for Cold-Start Micro-video Recommendation

TL;DR

This paper tackles cold-start micro-video recommendation by integrating a multi-view multimodal feature extractor with a hypergraph-based embedding layer and cross-view contrastive objectives. It models higher-order user–item interactions via a hypergraph and aligns multimodal representations through cross-modal and graph-hypergraph contrastive losses, trained alongside a BPR objective. Experiments on MicroLens-50K and MicroLens-100K show that MHCR consistently outperforms baselines, with notable gains in Recall@10 and NDCG@10 and substantial improvements under cold-start conditions. The results demonstrate that combining hypergraph signal propagation with self-supervision on multimodal data yields robust recommendations in sparse, cold-start environments.

Abstract

With the widespread use of mobile devices and the rapid growth of micro-video platforms such as TikTok and Kwai, the demand for personalized micro-video recommendation systems has significantly increased. Micro-videos typically contain diverse information, such as textual metadata, visual cues (e.g., cover images), and dynamic video content, significantly affecting user interaction and engagement patterns. However, most existing approaches often suffer from the problem of over-smoothing, which limits their ability to capture comprehensive interaction information effectively. Additionally, cold-start scenarios present ongoing challenges due to sparse interaction data and the underutilization of available interaction signals. To address these issues, we propose a Multi-view Hypergraph-based Contrastive learning model for cold-start micro-video Recommendation (MHCR). MHCR introduces a multi-view multimodal feature extraction layer to capture interaction signals from various perspectives and incorporates multi-view self-supervised learning tasks to provide additional supervisory signals. Through extensive experiments on two real-world datasets, we show that MHCR significantly outperforms existing video recommendation models and effectively mitigates cold-start challenges. Our code is available at https://github.com/sisuolv/MHCR.
Paper Structure (16 sections, 6 equations, 3 figures, 2 tables)

This paper contains 16 sections, 6 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. PROPOSED METHOD
  3. Problem Definition
  4. Multi-view Multimodal Feature Extraction Layer
  5. Multi-view Self-Supervised Learning
  6. Prediction
  7. EXPERIMENTS
  8. Experimental Settings
  9. Datasets
  10. Compared methods
  11. Evaluation Protocols
  12. Overall Performance (RQ1)
  13. Ablation Study(RQ2)
  14. Sensitivity Analysis (RQ3)
  15. Results on Cold-start Scenarios (RQ4)
  16. ...and 1 more sections

Figures (3)

  • Figure 1: Overview of the MHCR Architecture.
  • Figure 2: Ablation Study Results of MHCR.
  • Figure 3: Effect of Hypergraph Parameters and Contrastive Learning on Recommendation Performance.