论文部分内容阅读
以新浪微博为研究平台,在HITS(hyperlink-induced topic search)算法的基础上,提出融合用户交互行为和博文内容的微博用户可信度评估算法。分别构建基于交互行为和基于博文内容的微博用户有向链接图,图中节点表示用户,有向边体现用户基于交互行为或基于内容的指向关系;依据HITS算法计算两种拓扑结构下微博用户的权威度和中心度;以融合的权威度作为度量评估用户可信度。试验采用从新浪微博采集的数据作为测试集合,通过反复训练法获得可信度阈值,绘制不同可信度算法的用户可信度曲线,验证了算法的可行性和有效性。
Based on the HITS (hyperlink-induced topic search) algorithm, this paper proposes an algorithm to evaluate the credibility of Weibo users based on the interaction between Sina Weibo and Sina Weibo. Respectively, to construct a directed link graph of microblog users based on interaction behavior and blog content, in which nodes represent users and directed edges reflect interaction behaviors or content-based pointing relationships of users; and according to the HITS algorithm, calculate two kinds of microblogs User’s authority and degree of centrality; using the authority of fusion as a measure to evaluate user’s credibility. The test uses the data collected from Sina Weibo as a test set, obtains the credible threshold through iterative training, and draws the user credibility curve with different credibility algorithms to verify the feasibility and effectiveness of the algorithm.