论文部分内容阅读
微博作为一种新的社交网络,其影响力日益增大,逐渐成为用户获取第一消息的手段。为了增进微博服务,为用户提供兴趣信息,满足用户需求,本文综合考虑微博文本的特殊性,对原有的隐含狄利克雷分配(LDA)模型进行改进,提出基于User-topic的微博推荐模型。该模型在LDA的基础上添加了用户信息,从而得到主题生成用户的概率。通过微博文本-主题-词汇-用户之间的关系为用户提供多角度微博推荐,包括主题推荐、微博内容推荐和用户推荐。
As a new kind of social network, microblogging is increasingly influential and gradually becomes the means by which users get the first message. In order to improve the Weibo service and provide users with the interest information to meet the needs of users, this paper improves the original implicit Dirichlet allocation (LDA) model by considering the particularity of Weibo texts, Bo recommended model. The model adds user information based on LDA to get the probability that the topic generates users. Through the Weibo text - topic - vocabulary - the relationship between users to provide users with multi-angle Weibo recommendation, including the theme of recommendation, microblogging content recommendation and user recommendation.