Article Preview
Top1. Introduction
The rapid development of information technology has radically changed the way in which people communicate with each other. Mobile social networks (MSNs) play a crucial role in people’s daily communications, nowadays. A growing number of social network users with similar interests can converse and share information via their mobile devices (Ferreira et al., 2019; Su et al, 2018; Syn & Oh, 2015). According to the latest data from GSMA Intelligence, by the end of 2019, there were 5.19 billion unique mobile phone users in the world, up 100 million from the prior year.
Word-of-mouth communication is becoming much easier, and communication often goes viral in a short period, via simple clicks. However, fake news spread on social networks can be distinctively abrupt and extremely fast-spreading, which can adversely affect the normal social order (Chua & Banerjee, 2018). The term “online fake news” refers to “unconfirmed bits of information” in circulation (Zhao et al., 2016). When disasters or emergencies occur, online fake news can easily trigger public panic that could profoundly threaten social stability. People often share fake news in social groups in which the members are not necessarily familiar with each other. They forward fake news to other individuals or groups without verifying the credibility of the news. Research shows that fake news spreads “significantly farther, faster, deeper, and more broadly” than true news (Vosoughi, Roy, & Aral, 2018). Thus, there is an increasing interest in studying the spread of fake news on social networks (Xu et al., 2017).
Existing studies on fake news spreading largely focus on rumor identification, detection and prediction (Bondielli & Marcelloni, 2019; Xu et al., 2020; Yu et al., 2020; Zhang & Ghorbani, 2020). However, not many studies pay attention to the identification of influential users in the research area of fake news spreading. As fake news spreading is rapid and fluid, the identification and detection of fake news are generally conducted after fake news was spread to some extent. To reduce the spread of fake news on social network, one possible way is to identify those who play influential role in fake news spreading and educate them to check the credibility of news before disseminating news with others. If those influential people on the social network are able to change their behaviors to share or forward fake news, the severity of fake news spreading will be mitigated. To this end, we believe ranking the influence of users in the widespread of online fake news would be a useful technique to reduce future distribution of fake news.
News can reach an even greater number of people instantly (Gao et al., 2018) via MSNs, which usually consist of a huge number of users having something in common, such as careers, projects, common goals, or lifestyles. Influential nodes are those users with greater influence on social networks; their communications are more likely to infect others (Tulu et al., 2018). Given the existence of the internet’ s current dramatic progress, the ability to find the influential nodes on MSNs is becoming more and more critical.
In recent years, much attention has been given to identifying influential nodes on MSNs. A number of approaches and measures have been proposed from a variety of fields, including SNA (centrality analysis), clustering coefficient analysis, and eigenvector analysis (Chen et al., 2017; Wang et al., 2017; Wu et al., 2018). Degree centrality, betweenness centrality, and closeness centrality are the three basic measurements that are used to evaluate the influence of network nodes (Blaszczyk, 2018). The method of clustering coefficient analysis focuses on the network topological structure, the vertex’s local feature, and the connection of the set pair (Sisodia et al., 2016). The method of eigenvector analysis basically analyzes the importance of a network’s quality and the number of connections in the network.
A variety of evaluation models for ranking influential nodes have been proven to be efficient; however, an important question remains: which fundamental rules should the evaluation model follow? To answer this question, we present a novel model for evaluating the influence of individuals on social networks based on the theory of information entropy. The network influence of a node is a measurement of its uncertainty, which is the fundamental law of entropy. We construct a relationship graph to gain a basic understanding of the influence of nodes, and then we provide an innovative approach to ranking influential nodes to learn more about the spreading of fake news on MSNs. We present a model that measures the direct influence, indirect influence and comprehensive influence of nodes on MSNs in order to describe the complexity and the uncertainty of their influence.