Twitter knows: understanding the emergence of topics in social networks

Abstract

Social networks such as Twitter and Facebook are important and widely used communication environments that exhibit scale, complexity, node interaction, and emergent behavior. In this paper, we analyze emergent behavior in Twitter and propose a definition of emergent behavior focused on the pervasiveness of a topic within a community. We extend an existing stochastic model for user behavior, focusing on advocatefollower relationships. The new user posting model includes retweets, replies, and mentions as user responses. To capture emergence, we propose a RPBS (Rising, Plateau, Burst and Stabilization) topic pervasiveness model with a new metric that captures how frequent and in what form the community is talking about a particular topic. Our initial validation compares our model with four Twitter datasets. Our extensive experimental analysis allows us to explore several “what-if” scenarios with respect to topic and knowledge sharing, showing how a pervasive topic evolves given various popularity scenarios.