Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (1): 49-60.DOI: 10.3778/j.issn.1002-8331.2205-0314

• Research Hotspots and Reviews • Previous Articles     Next Articles

Survey on Knowledge Graph-Based Recommendation Methods

LUO Chengtian, YE Xia   

  1. Academy of Combat Support, Rocket Force University of Engineering, Xi'an 710025, China
  • Online:2023-01-01 Published:2023-01-01



  1. 火箭军工程大学 作战保障学院,西安 710025

Abstract: Recommendation methods are widely used in Internet scenarios to recommend personalized information for users, solve the problem of information overload, and improve user experience. Recommendation methods that introduce knowledge graphs can effectively solve the problems of data sparseness and cold start by using abundant side information, which contributes to the accuracy, diversity and explainability of recommendation, has aroused people’s research interest. This paper summarizes the existing recommendation methods and divides them into three types: embedding-based, path-based and propagation-based. It introduces and analyzes how existing methods mine entity and relationship information in knowledge graphs, and how to use knowledge graphs to make explainable recommendations, and compares the advantages and disadvantages of the three types of methods. The commonly used datasets in different application scenarios are introduced, and the challenging future research directions are prospected.

Key words: knowledge graph, recommendation method, explainable recommendation, graph embedding

摘要: 推荐算法广泛应用于互联网场景中,为用户推荐个性化的信息,解决信息过载问题,以提升用户体验。引入知识图谱的推荐算法利用丰富的辅助信息能有效解决数据稀疏和冷启动等问题,有助于推荐准确性、多样性和可解释性,引起了人们的研究兴趣。总结现有推荐方法,划分为基于嵌入、基于路径和基于传播三个类型,介绍分析现有方法如何挖掘知识图谱中的实体和关系的信息,以及如何利用知识图谱进行可解释的推荐,并对比了三种类型方法的优缺点;介绍了不同应用场景下的常用数据集;对具有挑战性的未来研究方向进行了展望。

关键词: 知识图谱, 推荐算法, 可解释推荐, 图嵌入