Dynamic SEIZ in online social networks: Epidemiological modeling of untrue information

Abstract

The epidemic propagation of untrue information in online social networks leads to potential damage to society. This phenomenon has attracted attention to researchers on a faster spread of false information. Epidemic models such as SI, SIS, SIR, developed to study the infection spread on social media. This paper uses SEIZ, an enhanced epidemic model classifies the overall population in four classes (i.e. Susceptible, Exposed, Infected, Skeptic). It uses probabilities of transition from one state to another state to characterize misinformation from actual information. It suffers from two limitations i.e. the rate of change of population and state transition probabilities considered constant for the entire period of observation. In this paper, a dynamic SEIZ computes the rate of change of population at fixed intervals and the predictions based on the new rates periodically. Research findings on Twitter data have indicated that this model gives more accuracy by early indications of being untrue information.