Characteristics of information diffusion in blogs, in relation to information source type

A novel method is presented to analyze the dynamics of social media, i.e., information diffusion properties, for information recommendation and ranking. In social media such as blogs, various information diffuses over time. As a result, a network structure is constructed. In an information diffusion network, each influential information source has an affected subnetwork whose nodes are reachable from it. We define three information diffusion properties of the subnetwork using the numbers of three types of directed two-edge connected subgraphs, which are basic structures in a directed acyclic graph such as an information diffusion network. Each basic structure type is related to information scattering, information gathering, or information transmission. We visualized and analyzed the structure of information diffusion networks extracted for various topics. Furthermore, we characterized the information diffusion properties by using the rank correlation coefficient, precision, and mean reciprocal rank and mean average precision of three types of information sources: official sites, news articles, and consumer generated media pages. We found that the three information diffusion properties have different characteristics and give priority to different types of information sources.