基于知识的推荐系统综述

A review of knowledge-based recommendation systems

基于知识的推荐算法在弥补过滤式传统推荐算法冷启动、数据缺失、“信息茧房”等问题上发挥着重要的作用。对基于知识的推荐算法相关研究进行评述,探讨其研究现状及研究进展。通过梳理CNKI相关文献和WOS中2017-2020的论文,采用内容分析法对国内外学术界基于知识的推荐算法的研究进展进行分析,对实际进展包括会话过程的优化、领域知识表示和获取、推理机制的发展、应用场景等进行总结和分析。基于知识的推荐算法与协同过滤相结合、与情感分析相融合将成为重要研究方向。

Knowledge-based recommendation algorithm plays an important role in compensating for the problems of filtering traditional recommendation algorithm such as cold start, data missing and "information cocoon". In this paper, the related research on knowledge-based recommendation algorithms is reviewed, and its development status and research progress are discussed. By combing CNKI related literatures and WOS papers from 2017 to 2020, content analysis was used to analyze the research progress of KBPR in domestic and foreign academic circles. The actual progress includes the optimization of the conversation process, the representation and acquisition of domain knowledge, the development of the reasoning mechanism, and the application scenarios are summarized and analyzed. The combination of collaborative filtering and sentiment analysis will become an important research direction.