基于文本语义和社会结构的可信度研究

Research on reliability based on text semantic and social structure

提出了一个基于文本上下文信息和用户社会结构的组合信任模型TCtrust,给出了置信模型,分别从显式和隐式信任对置信值进行了度量,设计出根据置信度排序的算法。通过对新浪微博热门话题"美国总统大选"进行实验,该算法在可信度上优于新浪微博的默认搜索算法,且能充分兼顾用户喜好、用户自身的影响力等因素,优于传统的LDA算法。

This paper proposed trust model based on the combination of text context information and the user social structure TCtrust, and gave confidence model, the confidence value from explicit and implicit trust measure, designed according to the confidence sort algorithms. By Sina microblogging hot topic＂ U. S. presidential election＂ experiment, the algorithm was superi- or in credibility the Sina microblogging default search algorithm, and could take into account user preferences, the user＇ s own influence factors such as superior to the traditional LDA algorithm.