Selective Intervention Strategy Based on Content Perception Model Against Fake News Sharing

This paper presents an intervention strategy, called a selective intervention, designed using a content perception model to analyze intervention effects at the cognitive level against fake news sharing. The content perception model derives the expected size of the suppression effect for each combination of content and user, so it allows selective intervention, that is, the selective use of the type of intervention from the perspective of maximizing the expected suppression effect. The model is developed as a Bayesian network, which assigns random variables to one of five layers: intervention, content feature, perceived feature, active state, and passive state. Here, two intervention types are introduced: accuracy-nudge based and correction based interventions. Based on a computer simulation technique, the effectiveness of selective intervention is compared with those of other simple intervention strategies in the context of political fake news studied in a previous work.