论文部分内容阅读
微博是Web2.0时代重要的网络服务工具,作为以用户为中心的信息发布、传播和分享平台,它包含了非常丰富的用户信息。在微博中,可以使用标签表示用户的兴趣和属性。而一个用户的兴趣和属性,通常包含在这个用户的文本信息和网络信息中。针对微博用户的标签进行分析,提出网络正则化的标签分发模型(NTDM)来为用户推荐标签。NTDM模型对用户个人简介中的词语和标签之间的关系进行建模,同时利用其社交网络结构作为模型的正则化因子。在真实数据集上的实验表明,NTDM在效果以及效率上都优于其他方法。
Weibo is an important Web service tool in the Web2.0 era. As a user-centric platform for information dissemination, dissemination and sharing, Weibo contains a wealth of user information. In Weibo, you can use tags to indicate the user’s interests and attributes. A user’s interests and attributes are usually included in the user’s text and network information. According to the analysis of the tags of Weibo users, a network regularized label distribution model (NTDM) is proposed to recommend tags for users. The NTDM model models the relationship between words and tags in a user’s profile, using their social network structure as a model regularization factor. Experiments on real datasets show that NTDM outperforms other methods both in performance and in efficiency.