基于用户上下文序列的个性化新闻推荐方法研究  

Research on News Recommendation Method Based on User Behavior Sequence

作　　者：温宇俊[1] 袁晖 WEN Yu-jun;YUAN Hui(Faculty of Science and Technology,Communication University of China,Beijing 100024,China;New Media Institute,Communication University of China,Beijing 100024,China)

机构地区：[1]中国传媒大学理工学部,北京100024 [2]中国传媒大学新媒体研究院,北京100024

出　　处：《中国传媒大学学报（自然科学版）》2018年第4期48-52,共5页Journal of Communication University of China：Science and Technology

基　　金：国家科技支撑计划(2014BAK10B01);中央高校基本科研业务费专项资金(3132016XNG1605)

摘　　要：面对互联网上丰富的新闻时,用户更希望阅读到自己感兴趣的那一部分新闻内容,避免太多资讯带来的信息过载问题,所以比搜索引擎等更加智能的个性化推荐技术成了新闻受众目前最迫切的需求之一。新闻推荐往往基于用户阅读记录实现,而当用户的阅读记录过少时,通常采用了流行度等简单方法实现,这类做法通常忽略了用户阅读的新闻序列之间其实还是有一定的关联,本文针对用户阅读过程中冷启动问题,从用户上下文序列的角度探讨其在新闻推荐中的影响。When facing a lot of news on the Internet,users may want to read some news that they are interested in,which can avoid information overload caused by too much information.So personalized recommendation technology,which is more intelligent than search engines,has become one of the most popular technologies for the news audience.The news recommendation often requires the user to read the record.Currently,when users have little reading history,system recommend news often by popularity factor or other simple method.In fact,it often ignores the connection between the series of news reading record.To solve the user cold-start problem,this paper discusses its influence on news recommendation from the perspective of user context sequence.

关 键 词：新闻推荐 记录序列 冷启动 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]