Social sensing model and analysis for event detection and estimation with Twitter

Abstract

Twitter is an Online Social Networking (OSN) service, in which users instantly share short messages with their followers. Active use of Twitter during major events turns it into a participatory sensor network, where each user acts as a sensor by sharing its observation through Twitter, and the observations may be utilized for estimation of various events. There are several studies on event detection in Twitter, which are primarily based on semantic analysis of collected tweets. To the best of our knowledge, there exists no analytical study on the accuracy of event detection and estimation of event signal with Twitter using observations collected. In this study, a communication theoretical model is developed for tweet propagation, and the accuracy of the estimated signal is explored with mean square error analysis. The results indicate that, in addition to user observations and network conditions, the accuracy of the estimation is also affected by user behavior patterns.