Fusing pattern discovery and visual analytics approaches in tweet propagation

Abstract

Abstract Over the past several years, social networks have become a major channel for information delivery. At present, social networks are being used to obtain more followers and exert influence over people during political campaigns. However, the propagation of a social network post is dependent on numerous factors. Some of these are known; for example, the post contents, the time when it was posted, and the person or entity by whom it was posted. However, other factors remain unknown, such as what makes a post more successful than others, and how posts from similar profiles evolve and propagate differently over time. The main subject of this work is addressing these types of questions. Our approach relies on a three-fold methodology for studying the influence and propagation of posts: graph-based, semantic, and contrast pattern recognition analysis. The results obtained are complemented by a dynamic visualization that encompasses all of the variables involved. In order to corroborate our results, we collected all posts from the Twitter accounts of the most prominent Mexican political figures and analyzed the influence and propagation of each post issued.