Top-N Collaborative Filtering Recommendation Algorithm Based on Knowledge Graph Embedding

• Published in: Knowledge Management in Organizations Vol. 1027; pp. 122 - 134
• Main Authors: Zhu, Ming, Zhen, De-sheng, Tao, Ran, Shi, You-qun, Feng, Xiang-yang, Wang, Qian
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2019; Springer International Publishing
• Series: Communications in Computer and Information Science
• Subjects: Collaborative filtering; Knowledge graph embedding; Recommendation algorithm; Semantic similarity
• ISBN: 3030214508; 9783030214500
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-3-030-21451-7_11

The traditional collaborative filtering recommendation algorithm only uses the item-user rating matrix without considering the semantic information of the item itself, resulting in a problem that the recommendation accuracy is not high. This paper proposes a Top-N collaborative filtering recommendation algorithm based on knowledge graph embedding. The knowledge graph embedding is used to learn a low-dimensional vector for each entity and relationship in the knowledge graph, while maintaining the structure and semantic information of the original graph in the vector. By calculating the semantic similarity between items, the semantic information of the item itself is incorporated into the collaborative filtering recommendation. The algorithm makes up for the defect that the collaborative filtering recommendation algorithm does not consider the knowledge information of the item itself, and enhances the effect of collaborative filtering recommendation on the semantic level. The experimental results on the MovieLens dataset show that the algorithm can get higher values on precision, recall and F1 measure.