结合微博关注特性的UF_AT模型用户兴趣挖掘研究.pdfVIP

优先出版 计 算 机 应 用 研 究 第32 卷 结合微博关注特性的UF_AT 模型用户兴趣挖掘研究 1 1 2 1 王永贵 ，张 旭 ，任俊阳 ，刘宪国 (1．辽宁工程技术大学 软件学院，辽宁 葫芦岛 125105；2 ．吉林省煤业集团有限公司，长春 130012) 摘 要：微博作为国内主流社交网站，信息量与日俱增。目前微博用户兴趣挖掘方法多停留在研究用户浏览网页时点击 行为、用户所发微博内容或所在社区等表象层面，尚未深入到微博用户使用特性层面。从用户微博内容出发，结合用户 关注对象微博，提出一种改进作者主题模型UF_AT （users focus-author topic ）。最后对真实数据进行实验得出，模型在 用户兴趣主题以及主题词概率值上均高于AT 模型，而且用户兴趣主题准确、全面，同时验证了UF_AT 模型在挖掘用 户兴趣中的有效性。 关键词：微博；用户关注特性；作者主题模型；兴趣挖掘； 中图分类号： TP181 Research on micro-blog user ’s interest mining based on UF_AT model which combining with focusing feature of microblog 1 1 2 1 WANG Yong-gui , ZHANG Xu , REN Jun-yang , LIU Xian-guo (1. College of software, Liaoning Technical University, Liaoning, , Huludao, 125105, China; 2. Jilin Coal Industry Group Co. , Ltd. , Jilin, , Changchun, 130012, China) Abstract Microblog as a mainstream social networking sites, The information is increasing by the explosive growth. Currently the methods of microblog user interest mining most remain in the level such as appearance of the research of clicking action when the user browsing websites, the microblog content user post or the community they belong, yet in-depth features to microblog user level. From the content of users posting , combined with the focused users’microblog, proposing an improved AT model UF_AT. At last the experiments on the real data show that the probability values of user ’interest topics and the words on the topics of the model are higher than the AT