Abstract: With the increasing number of users, the credibility crisis arises on micro-blog. This paper defines a novel concept called user credibility and presents an evaluation model called User-Rank to measure the credibility of a registered user. The User-Rank model consists of a self-scoring model and a mutual-scoring model. Then this paper proposes an algorithm to compute and rank all users’ credibility scores. The experimental results on datasets extracted from Sina Weibo show that the proposed algorithm outperforms the baseline algorithms, and its time complexity only reaches O(n²).

Key words: user credibility, Sina Weibo, User-Rank

摘要： 随着新浪微博用户数的增加，微博上的可信度危机凸显出来。定义了一种叫做用户可信度的新概念,并提出了一种评估用户可信度的模型User-Rank。该模型由自评估模型和互评估模型两个子模型组成。提出了一种计算用户可信度，并对其得分进行排序的算法。在基于新浪微博平台抽取的数据集上进行了实验，结果表明该算法优于一般的基础性算法，且互评估模型算法的时间复杂度仅为O(n²)。

关键词: 用户可信度, 新浪微博, User-Rank