DiffuScope: inferring post-specific diffusion network

Abstract

Post-specific diffusion network elucidates the who-saw-from-whom paths of a post on social media. A diffusion network for a specific post can reveal trustworthy and/or incentivized connections among users. Unfortunately, such a network is not observable from available information from social media platforms; hence an inference mechanism is needed. In this paper, we propose an algorithm to infer the diffusion network of a post, exploiting temporal, textual, and network modalities. The proposed algorithm identifies the maximum likely diffusion network using a conditional point process. The algorithm can scale up to thousands of shares from a single post and can be implemented as a real-time analytical tool. We analyze inferred diffusion networks and show discernible differences in information diffusion within various user groups (i.e. verified vs. unverified, conservative vs. liberal) and across local communities (political, entrepreneurial, etc.). We discover differences in inferred networks showing disproportionate presence of automated bots, a potential way to measure the true impact of a post.