LLM-Guided Multi-View Hypergraph Learning for Human-Centric Explainable Recommendation This paper proposes a novel explainable recommendation framework, LLMHG, which integrates the reasoning capabilities of large language models (LLMs) with the structural advantages of hypergraph neural networks. The framework aims to enhance recommendation systems by capturing the multifaceted nature of human interests through semantic reasoning and hypergraph structure learning. LLMHG extracts Interest Angles (IAs) from user behavior data, which are structured representations of user preferences. These IAs are used to construct a multi-view hypergraph, which is then refined through hypergraph structure learning to focus on the most salient aspects of user preferences. The refined hypergraph is integrated with latent embeddings from a sequential recommendation model to generate recommendations. The framework is validated on real-world datasets, showing that LLMHG outperforms conventional models in terms of recommendation performance and explainability. The proposed method offers a pathway to apply advanced LLMs for better capturing the complexity of human interests across machine learning applications. The framework is evaluated on three benchmark datasets: Amazon Beauty, Amazon Toys, and MovieLens-1M. The results show that LLMHG significantly improves performance metrics such as HR@n and NDCG@n. The method is also compared with other baselines and shows that it outperforms them in terms of performance and cost-effectiveness. The paper also discusses the limitations of the current approach and suggests future directions for improvement.LLM-Guided Multi-View Hypergraph Learning for Human-Centric Explainable Recommendation This paper proposes a novel explainable recommendation framework, LLMHG, which integrates the reasoning capabilities of large language models (LLMs) with the structural advantages of hypergraph neural networks. The framework aims to enhance recommendation systems by capturing the multifaceted nature of human interests through semantic reasoning and hypergraph structure learning. LLMHG extracts Interest Angles (IAs) from user behavior data, which are structured representations of user preferences. These IAs are used to construct a multi-view hypergraph, which is then refined through hypergraph structure learning to focus on the most salient aspects of user preferences. The refined hypergraph is integrated with latent embeddings from a sequential recommendation model to generate recommendations. The framework is validated on real-world datasets, showing that LLMHG outperforms conventional models in terms of recommendation performance and explainability. The proposed method offers a pathway to apply advanced LLMs for better capturing the complexity of human interests across machine learning applications. The framework is evaluated on three benchmark datasets: Amazon Beauty, Amazon Toys, and MovieLens-1M. The results show that LLMHG significantly improves performance metrics such as HR@n and NDCG@n. The method is also compared with other baselines and shows that it outperforms them in terms of performance and cost-effectiveness. The paper also discusses the limitations of the current approach and suggests future directions for improvement.