Mitigating Misinformation in Online Social Network with Top-k Debunkers and Evolving User Opinions

Co-spread of Misinformation and Fact-Checking Content During the Covid-19 Pandemic

“…This echoes previous work that describes the enticement of emotion and novelty in misinformation [32]. Likewise, this also indicates that the communities sharing fact-checks and those sharing misinformation are likely different indicating that previous agent-based models that address the impact of factcheckers on a network [28,23] may need to be adjusted for lower-than-expected inter-community contact. Finally, significant differences in sharing behaviour appears mostly during the ramping up period of the pandemic (early phase) with large variations in deviation and means toward misinformation.…”

“…Most studies of misinformation spread tend to be focused on early intervention and removal [3]. In this context, many works have focused on the application and extensions of epidemiological models [12,13,3] with additional features like weighted values for particular users [29] or, more recently, information about debunkers and the dynamics of opinion evolution [23]. Notably, Saxena et al [23] demonstrated that identifying influential nodes may also help exploit the spread of fact-checked information and impact user opinion over time.…”

Identifying Cost-effective Debunkers for Multi-stage Fake News Mitigation Campaigns

“…One class of studies focus on selecting debunkers to maximize the spread of truthful information to counteract the fake news spread on social networks. Some studies heuristically select top 𝑘 most influential users as debunkers [18,19]. Their assumption is that users with high social influence produce wide propagation of true news on social networks.…”

“…2 Based on the current propagation state of fake and true news on social networks, our solution dynamically selects debunkers within budget at each stage with the objective to maximize the cumulative mitigation effect -more true news will be propagated to users exposed to more fake news -across stages of the campaign. Different from studies by Saxena [18,19], our selected debunkers are not necessarily the most influential users on social networks but only those with maximum influence over the users who have been exposed to fake news. Different from the study by Farajtabar [5], our debunkers are dynamically selected for each stage according to the fake news propagation state at the time, that is, the set of debunkers may differ from stage to stage.…”

“…There have been limited studies on automatic fake news mitigation. Some studies heuristically select users of high social influence -having a large number of followers -as debunkers [18,19] to propagate true news to mitigate the spread of fake news. The assumption is that influential users on the social network produce wide propagation of true news.…”