In this paper, we use graph theory to explore concepts of influence in socialized groups. When analyzing social networks, centrality indicators make it possible to assess the power of an individual. We discuss various centrality indicators and focus on degree and betweenness. After observing a strong correlation between them, we propose defining new assessments based on a decorrelation method that we characterize from different mathematical perspectives (algebraic, probabilistic, and statistical). We apply this theoretical framework to a network of tweets about the Uber versus taxi conflict, which took place in June, 2015, and for which we detected different influential individuals.
Boulet, R., & Lebraty, J. (2018). A New Influence Measure Based on Graph Centralities and Social Network Behavior Applied to Twitter Data. Communications of the Association for Information Systems, 43, pp-pp. https://doi.org/10.17705/1CAIS.04321