Measuring time-sensitive user influence in Twitter

Identification of the influential users is one of the most practical analyses in social networks. The importance of this analysis stems from the fact that such users can affect their followers "/friends" viewpoints. This study aims at introducing two new indices to identify the most influential users in the Twitter social network. Four sets of features extracted from user activities, user profile, tweets, and actions performed on tweets are deployed to create the proposed indices. The available methods of detecting the most influential Twitterers either consider a limited set of features or do not accurately measure the effect of each feature. The indices proposed in this paper consider a comprehensive set of features and also provide a time-sensitive rank which can be used to measure the dynamic nature of influence. Moreover, the relative impact of each feature is computed and considered in the indices. We employ the indices to discover the influential Twitter users posting on Paris attacks in 2015, in a comprehensive analysis. The influence trend of users' tweets in a 21-day period discloses that 76% of the users do not succeed in posting a second influential tweet. Results reveal that the proposed indices can detect both the publicly recognized sources (like celebrities) and also the less known individuals which gain credit by posting several influential tweets after a specific event. We further compare the proposed indices with other available approaches.