Learning influence probabilities in social networks

by Goyal et al. (goya2010)

The article's authors learn influence models based on

• social graphs and
• an action log

Based on the learned models they are able to predict i) whether a user will perform a specific action and ii) when it will be performed.

A user's properties are encoded in a user influenceability and an action influence quotient. According to the authors were Domingos and Richardson (domingos2001, richardson2002) the first to consider the propagation of influential users from a data mining perspective.

Related Literature:

• information diffusion
• user influence (social networks)
• outbreak detection of viri (leskovec2007a)
• diffusion of innovations

Tang et. al (tang2009) consider topic-specific influence analysis by identifying topic-specific subnetworks and influence weights. They discuss applying their approach to the expert finding problem.