中圖分類號(hào)：TN0 文獻(xiàn)標(biāo)志碼：A DOI: 10.16157/j.issn.0258-7998.256532
中文引用格式： 康永玲. 基于知識(shí)圖譜和協(xié)同過(guò)濾算法的多頭注意力網(wǎng)絡(luò)[J]. 電子技術(shù)應(yīng)用，2025，51(8)：60-64.
英文引用格式： Kang Yongling. Multi-head attention network based on knowledge graph and collaborative filtering algorithm[J]. Application of Electronic Technique，2025，51(8)：60-64.

Multi-head attention network based on knowledge graph and collaborative filtering algorithm

Abstract： Most current recommendation methods based on knowledge graphs focus on the encoding mechanism of knowledge associations, often neglecting the potential key collaborative signals in user-item interactions. This leads to the learned embedding vectors of existing models being unable to effectively represent the latent semantics of users and items in the vector space. To address this issue, this paper proposes a multi-head attention network that integrates knowledge graphs and collaborative filtering - the collaborative knowledge-aware multi-head attention network (CKAN-MH). This network introduces a multi-head attention mechanism on the basis of the traditional CKAN model to adaptively focus on different subsets of features and perform differential weighting of tail entities by dynamically adjusting attention weights. After introducing the multi-head attention mechanism, the model can more comprehensively capture the complex relationships and patterns hidden in the data, thereby significantly improving the performance of the recommendation system. Additionally, we conducted experimental evaluations on three real datasets using the CKAN-MH model. The experimental results show that the CKAN-MH model outperforms several current mainstream advanced baseline models in terms of performance, verifying the effectiveness and superiority of this model.

Key words : recommendation system；knowledge graph；collaborative filtering；multi-head attention network