How to differentiate propagators of information and misinformation–Insights from social media analytics based on bio-inspired computing

Abstract

People react to informational content in social media and often propagate the same. The content available may either be authentic information or misinformation. This study focuses on modeling the user attributes of information and misinformation propagators, derived from user based and user generated content based attributes. In this study, 10,000 users and 5,55,684 tweets were analyzed to compute eighteen factors based on tweet and user parameters. Factor selection for the final analysis identified 11 statistically relevant factors. An approach is proposed to classify users as information or misinformation propagators using K-means integrated with bio inspired algorithms like firefly, cuckoo search and bat algorithms. Results show that the firefly algorithm with levy flights gives the highest accuracy while the bat algorithm converges to an optimum solution faster. Findings indicate that factors like emotion stability, polarity stability, hashtag consolidation ratio, hashtag diversity, lexical diversity, favorites count and friends count have relatively higher importance in predicting propagators differently. Computational findings are integrated with psychological behaviors of people by building upon the theory of personality traits. Findings are useful in domains of viral marketing and information governance to identify potential user groups which may play a role in the propagation of misinformation and information.

Subject Classification:

• 62H30
• 68T50
• 92B20

Keywords:

• Social media analytics
• Content virality
• Twitter analytics
• Information diffusion
• Bio-inspired computing
• Machine learning