Personalized service recommendation for smart hotels based on knowledge graph embeddings

Computer-based recommendation systems play a crucial role in enhancing user experiences within smart hotel environments by providing personalized services. Traditional approaches, such as collaborative filtering and deep learning, often fail to address challenges associated with data sparsity and the modeling of complex semantic and contextual relationships. To overcome these limitations, this study proposes a knowledge graph embedding-based framework for personalized service recommendation. By leveraging advanced embedding techniques, including TransE, DistMult, and RotatE, the framework captures multirelational semantic patterns and integrates contextual information, such as temporal and spatial factors, to improve recommendation quality. The proposed framework was evaluated on a hotel recommendation dataset and demonstrated significant improvements in accuracy, diversity, and coverage compared to baseline models. The use of a re-ranking strategy further enhanced the diversity of recommendations while maintaining high relevance, ensuring a balance between personalization and variety. Key contributions of this work include the integration of knowledge graph embeddings with contextual modeling to address data sparsity and the development of an efficient embedding-based training pipeline. These findings highlight the effectiveness of knowledge graph embeddings in building scalable and accurate recommendation systems for smart hotels.