基于改进HITS算法的微博用户影响力计算方法  

Calculating Microbloggers’Influence Based on Improved HITS Algorithm

作　　者：齐佳 徐建民 Qi Jia;Xu Jianmin(School of Cyber Security and Computer,Hebei University)

机构地区：[1]河北大学网络空间安全与计算机学院,河北保定071002

出　　处：《图书馆杂志》2022年第4期106-114,共9页Library Journal

基　　金：国家社会科学基金后期资助项目“基于术语关系的贝叶斯网络检索模型扩展”(项目编号:17FTQ002)的研究成果之一。

摘　　要：针对传统微博网络用户影响力度量方法中用户关系不明确、时间特征不明显的问题,从网络结构、交互行为和时间3个角度考虑影响力的个体差异,提出一种基于改进HITS算法的微博用户影响力计算方法。该方法以用户间动态交互行为作为切入点,选择其动态交互强度作为改进算法的边权重,通过sigmoid函数对中心值进行修正,整合用户的权威值和中心值量化用户影响力,提高算法的适应性,计算结果更为合理,为微博用户影响力计算提供了新的思路。以新浪微博为实验对象,实验结果表明,该方法相对于传统影响力计算方法,在DCG值、查准率、召回率和F值评价指标方面均有所提升。To address the problems in the traditional method of measuring microblogger’s influence,such as the unclear user relationship and unclear time characteristics,a modified HITS algorithm is proposed based on the differences in influence from the perspectives of network structure,interaction behavior,and time.It selects the dynamic interaction intensity between users as the edge weight,revises the central value through the sigmoid function,and integrates users’authoritative value and the central value to quantify their influence.As a result,the algorithm has increased adaptability,the results are more reasonable.It then tested the algorithm using Sina microbloggers.Compared with the traditional method of calculation,the new does improve in terms of the DCG value,precision rate,recall rate and F-score evaluation index.

关 键 词：HITS算法 微博用户 影响力 动态交互 时间片 

分 类 号：TP393.092[自动化与计算机技术—计算机应用技术] TP311.13[自动化与计算机技术—计算机科学与技术]