推荐系统用户模型的研究热点及启示——基于近十年核心文献的知识图谱分析

Hot Topics and Enlightenment of User Model in Recommendation System:Analysis of Knowledge Graph Based on Core Literatures in Recent Ten Years

【目的/意义】为了探析推荐系统中用户模型的研究热点,本文以近十年国内208篇核心文献为研究样本,系统、规范地分析了关于推荐系统用户模型的研究近况,以期为推荐系统用户建模的未来研究方向提供参考。【方法/过程】文章分别利用SATI、Ucinet和SPSS软件进行高频关键词共现、知识图谱和聚类分析。【结果/结论】高频关键词的聚类分析结果显示,我国推荐系统用户模型研究的热点主要有4个分支主题:模型表示、模型学习、模型更新、模型应用;研究启示主要有:加强用户模型表示方法的组合使用,促进综合模型构建;融合多场景、跨系统更新用户模型,实现用户模型的复用;强化用户模型的个性化应用,提高用户个性化推荐服务质量;注重用户模型多维度研究,提高资源精准推荐效果。

【Purpose/significance】In order to explore the hotspots of user model research of recommendation system,this paper takes 208 domestic core literatures in recent ten years as research samples,systematically and normatively analyzes the recent research situation of recommendation system user model,in the hope of providing references for the future research direction of recommendation system user modeling.【Method/process】In this paper,SATI,Ucinet and SPSS are used to analyze the co-occurrence,knowledge map and clustering of high-frequency keywords.【Result/conclusion】The results of clustering analysis of high-frequency keywords show that the research hotspot of recommendation system user model in China mainly includes four sub-topics:model representation,model learning,model updating,model application.The research enlightenment mainly includes the following aspects:researchers should enhance the combination of user model representation methods,and promote the construction of comprehensive model;they should integrate multiple scenarios,update user models across systems,and reuse user models;they should strengthen the personalized application of user model,and improve the service quality of personalized recommendation;they should pay attention to multi-dimension research of user model,and improve the accurate recommendation effect of resources.