Identification of opinion leader on rumor spreading in online social network Twitter using edge weighting and centrality measure weighting

Abstract

Rumor spreading has been an essential issue for society. One of the platforms for rumor spreading is Twitter. Finding the opinion leader of its issue is also important in order to know who are users whom have a high impact of bringing the issue. So we can give the suggestion to the law authority to give the right authorization afterward. Opinion leader can be found using centrality measure metric on social network analysis study. This study has node and edge as its property. For recent years, there are many conducted researches about centrality measure. Some of them are combining some centrality measures together. Aside from defining the centrality measure, defining the edge is also important. Twitter has a different kind of relationships that can be turned into an edge, but not all the relationships have the same impact for spreading the rumor. This study conduct two experiments, first experiment is edge weighting. This experiment is aimed to see the importance of each defined edge type for finding the opinion leader. The second experiment is centrality weighting. This experiment is aimed to see the weight that could give more accurate opinion leader based on other evaluation algorithms. The study found the edge that has the ability to spread to wider audience (quote, retweet, and reply) tend to have a bigger impact for finding opinion leader than mention relationship. The study also finds that a low in-degree weight, high betweenness weight, and low or no PageRank weight could give 100% agreement upon other evaluation algorithms for finding the opinion leader.