Modeling Information Diffusion in Implicit Networks

摘要: Social media forms a central domain for the production and dissemination of real-time information. Even though such flows information have traditionally been thought as diffusion processes over social networks, underlying phenomena are result complex web interactions among numerous participants. Here we develop Linear Influence Model where rather than requiring knowledge network then modeling by predicting which node will influence other nodes in network, focus on global rate through (implicit) network. We model number newly infected function got past. For each estimate an that quantifies how many subsequent infections can be attributed to time. A nonparametric formulation leads simple least squares problem solved large datasets. validate our set 500 million tweets 170 news articles blog posts. show accurately models influences reliably predicts temporal dynamics diffusion. find patterns individual participants differ significantly depending type topic