Network Analysis of Three Twitter Functions: Favorite, Follow and Mention

We analyzed three functions of Twitter (Favorite, Follow and Mention) from network structural point of view. These three functions are characterized by difference and similarity in various measures defined in directed graphs. Favorite function can be viewed by three different graph representations: a simple graph, a multigraph and a bipartite graph, Follow function by one graph representation: a simple graph, and Mention function by two graph representations: a simple graph and a multigraph. We created these graphs from three real world twitter data and found salient features characterizing these functions. Major findings are a very large connected component for Favorite and Follow functions, scale-free property in degree distribution and predominant mutual links in certain network motifs for all three functions, freaks in Gini coefficient and two clusters of popular users for Favorites function, and a structure difference in high degree nodes between Favorite and Mention functions characterizing that Favorite operation is much easier than Mention operation. These finding will be useful in building a preference model of Twitter users.