融合用户背景和用户人格的话题推荐方法  

A TOPIC RECOMMENDATION METHOD INTEGRATING USER'S BACKGROUND AND PERSONALITY

作　　者：范洪博[1] 杨笑锋 张晶[1] Fan Hongbo,Yang Xiaofeng,Zhang Jing(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,Yunnan,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《计算机应用与软件》2018年第7期309-312,333,共5页Computer Applications and Software

摘　　要：话题是用户兴趣的表现形式之一,目前对话题的推荐方法主要是根据用户最新发布的文本信息进行协同过滤推荐,但其存在着推荐话题单一且重复,无法对用户进行多兴趣话题推荐的问题。张磊等提出一种基于用户文本信息集对用户进行人格预测的方法对用户话题推荐,虽然解决了推荐话题单一和重复的问题,但仍存在着推荐精准度不高的缺点。针对上述问题,提出一种融合用户背景信息和用户人格信息进行话题推荐的方法,根据用户发布的文本信息集对用户进行人格特质预测,根据用户的人格特质分析用户的潜在兴趣点,结合背景信息和潜在兴趣点,利用协同过滤算法对进行用户话题推荐。实验结果表明,该方法对用户话题推荐的精准度有所提高。其推荐精准度相较于基于用户间背景相似的文本推荐算法(BG-CF)提升19.74%,相较于人格特质相似的文本推荐算法(CS-CF)算法提升8.92%。The topic is one of the expressions of the user＇s interests. At present,the topic recommendation method is mainly based on the user＇s latest published text information for collaborative filtering recommendation. However,there is a problem that the topics of recommendation are single and repetitive, and the multi-interest topics cannot be recommended to users. Zhang Lei proposed a method to recommend user ＇s topics according to the user ＇s text information set to predict the user ＇s personality. Although this method can solve the above problem,it still has shortcoming of low precision. Aiming at the above problems,this paper puts forward a method combining user ＇s background information and personality information for topic recommendation. The personality traits of users are predicted according to the user＇s published text information set. And the potential points of interest of users are analyzed based on the personality traits of users. Combining with background information and potential points of interest,collaborative filtering algorithm is applied for user topic recommendations. The experimental results show that the proposed method has improved the accuracy of the user topic recommendation. Its recommendation accuracy is improved by 19. 74% compared to the text recommendation algorithm based on user background similarity（ BG-CF）,and is also improved by 8. 92% compared to the text recommendation algorithm（ CS-CF） with similar personality traits.

关 键 词：话题推荐 用户人格 背景信息 协同过滤 MAE 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]