摘要
[研究目的]准确识别网络舆情传播源点成为提升网络舆情治理能力和创新社会治理的目标。由于单一中心性不能较好地利用网络数据集的结构特征对舆情信息源点进行追溯定位,降低了溯源准确率。因此,提出一种基于多中心性分析的网络舆情信息源点追溯算法。[研究方法]首先,在随机图与正则图上进行舆情传播与溯源仿真实验;其次,将该算法与现有溯源算法的性能进行对比,进一步在真实网络——美国电网上验证该算法的有效性;最后,分析影响舆情信息源点追溯准确率的潜在因素。[研究结论]研究结果表明,该算法在随机图与正则图上的溯源准确率较单一中心性算法的溯源准确率提升约83%与41%,溯源平均错误距离仅为0.02步和0.5步,且在真实网络上的溯源平均错误距离为6.56步,优于动态年龄算法(DA)的11.3步与最小描述长度算法(MDL)的8.88步。
[Research purpose]It has become the goal of improving capabilities of the governing online public opinion and innovating social governance that accurately identify the source of the spread of online public opinion.Since the single centrality can not use the structural characteristics of the network data set to trace and position the source of public opinion information,which reduces the traceable accuracy.this paper proposes an algorithm for tracing the source of information of online public opinion based on the multicentricity analysis.[Research method] This paper conducts simulation experiments of public opinion propagation and traceability on random graphs and regular graphs firstly,and then compares the performance of the algorithm with the existing traceability algorithms,further verifies the effectiveness of the algorithm on the real network—the U.S.Power Grid,and analyzes the potential factors that affect the traceable accuracy of source points of public opinion information.[Research conclusion] The research results show that the traceable accuracy of the proposed algorithm on random graphs and regular graphs is about 83% and 41% higher than that of single centrality algorithms relatively,the traceable average error distance is only 0.02 steps and 0.5 steps relatively,and the traceable average error distance on the real networks is 6.56 steps,which is better than 11.3 steps calculated by using the dynamic age algorithm(DA) and 8.88 steps calculated by using the minimum description length algorithm(MDL).
作者
于凯
白西柯
郭煜婕
Yu Kai;Bai Xike;Guo Yujie(School of Public Administration,Xinjiang University of Finance and Economics,Urumqi 830012;School of Information Management,Xinjiang University of Finance and Economics,Urumqi 830012)
出处
《情报杂志》
CSSCI
北大核心
2022年第3期166-172,共7页
Journal of Intelligence
基金
新疆维吾尔自治区自然科学基金项目“基于多层网络模型的信息传播源头定位研究”(编号:2019D01A22)
新疆维吾尔自治区天山青年计划项目“在线社交网络信息传播机制与链路预测研究”(编号:2018Q027)
新疆财经大学研究生科研创新项目“基于修正的SEIR新冠肺炎传播预测及模型评估分析”(编号:XJUFE2020K042)项目成果之一。
关键词
多中心性
网络舆情
信息源点
源点追溯
multicentricity
network public opinion
information source point
source point tracing
