微博个性化标签图形化RTM模型Gibbs采样推荐

发布时间：2018-04-19 19:41

本文选题：Gibbs采样 + 微博标签　；参考：《微电子学与计算机》2017年12期

【摘要】：为提高个性化标签推荐方法性能,提出基于Gibbs采样推理的微博个性化标签隐含关系主题模型(Relation Topic Model,RTM)推荐算法.首先,利用图形化形式对微博中的潜在局部信息进行表达,对用户主题分布为代表的用户进行top-k相似用户发现,然后计算出现在这些用户中的所有标签的频率,并推荐与用户最相关的标签.其次,为挖掘潜在主题信息,利用带惩罚项的增强型余弦相似度RTM模型对微博标签进行命名,大大提高联合建模对潜在主题生成标签的影响,并可发现全局标签和主题之间的关系;最后,通过真实的实验结果显示,所提推荐方法要优于选取的TF-IDF、RTMSA等几种经典标签推荐算法,验证了算法有效性.
[Abstract]:In order to improve the performance of personalized tag recommendation method, a recommendation algorithm based on Gibbs sampling reasoning for Weibo personalized tag implied relation Topic model is proposed. First, the potential local information in Weibo is expressed graphically, and the users whose themes are distributed are found by top-k similar users, and then the frequency of all tags appearing in these users is calculated. And recommend the most relevant label to the user. Secondly, in order to mine the potential topic information, the enhanced cosine similarity RTM model with penalty term is used to name Weibo tag, which greatly improves the influence of joint modeling on the potential theme generation tag. The relationship between the global tag and the topic can be found. Finally, the experimental results show that the proposed recommendation method is superior to several classical tag recommendation algorithms such as TF-IDF / RTMSA, which verifies the effectiveness of the proposed algorithm.
【作者单位】：武汉理工大学计算机科学与技术学院;
【分类号】：TP391.3

【相似文献】