跨领域推荐研究进展

Research Progress of Cross-domain Recommendation

数据稀疏性和冷启动一直是当前个性化推荐面临的主要问题,跨领域推荐因其可利用多源多领域数据,为解决这两个问题提供了方法与思路。跨领域推荐的核心在于辅助领域数据的利用和知识的迁移。本文从单辅助域跨领域推荐和多辅助域跨领域推荐两个角度,对当前国内外跨领域推荐技术进行了系统的研究与分析,阐述了现有单辅助域跨领域推荐和多辅助域跨领域推荐中使用的跨领域推荐方法,同时讨论了已有方法各自的优缺点与适用范围,并指出了跨领域推荐研究的未来研究方向。

Data sparsity and cold start are always the main problems faced by personalized recommendation.Cross-domain recommendation provides methods and ideas to solve these two problems because it can use multi-source and multi-domain data.The core of cross-domain recommendation is the use of source-domain data and knowledge migration.This paper systematically studies and analyses the current research situations of the cross-domain recommendation technologies and from the perspectives of single-source cross-domain recommendation and multi-source cross-domain recommendation,expounds the existing cross-domain recommendation methods used in single-source cross-domain recommendation and multi-source cross-domain recommendation,and discusses the advantages and disadvantages of the existing methods and their scope of application.Finally,pointed out the future research directions of the cross-domain recommendation techniques.