摘要
随着Web 2.0技术的发展,社会网络为人们进行交流和协作提供了新的便捷平台。面对网络信息过载问题,在海量的信息中找到自己感兴趣并信任的领域专家,参考专家意见做抉择,变得十分困难。本文提出一种基于信任的LDA(latent Dirichlet allocation)主题模型社会网络中领域专家推荐方法,实现了基于用户信任的领域专家个性化推荐。该方法以LDA主题模型为基础,综合考虑社会网络结构、用户间的信任关系及社会影响力,弥补了传统专家推荐方法只考虑专家特征,导致专家推荐精度不高及推荐结果模式化的不足。最后通过实验验证了该方法的可行性和有效性。
With the development of Web 2.0 technology,the social network has provided a new and convenient platform to communicate and collaborate for people.Faced with the problem of network information overloaded,it is very difficult to find the experts who are interested in the topic and are reliable to make decisions based on experts’opinions.In this paper,domain experts recommendation in social network based on the LDA theme model of trust method is proposed.The personalized recommendation of domain experts for the user trust is implemented.The method is based on the topic model(LDA),the social network structure,the users’trust relationship and social influence.It makes up the deficiency of the traditional experts recommendation methods that only involve expert’s characteristics which lead to low recommendation accuracy and the problem of the pattern of recommended results.Finally,the feasibility and effectiveness of the method are verified by experiments.
作者
刘电霆
吴丽娜
LIU Dianting;WU Lina(College of Mechanical and Control Engineering,Guilin University of Technology,Guilin Guangxi 541004,China;College of Information Science and Engineering,Guilin University of Technology,Guilin Guangxi 541004,China)
出处
《广西师范大学学报（自然科学版）》
CAS
北大核心
2018年第4期51-58,共8页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家自然科学基金(51165004)
广西科学研究与技术开发计划(桂科攻1598007-15)
关键词
社会网络
信息过载
主题模型
个性化推荐
用户信任
social network
information overload
theme model
personalized recommendation
user trust
