The impact of individual attributes on knowledge diffusion in web forums

Web forums one of the most important Web 2.0 developments have become a useful platform for tracking and exploring knowledge diffusion among online users. Such networks are usually large, complex, and multi-dimensional, consisting of various types of nodes and ties. Therefore we use exponential random graph models to model online multi-dimensional knowledge diffusion and to systematically evaluate the impact of individual characteristics on knowledge diffusion networks. Experiments are conducted on a longitudinal dataset that covers one decade, drawn from the Yahoo! Finance Wal-Mart message board. The results show that reciprocal knowledge diffusion occurs in the forum; high-authority individuals play an important role in knowledge diffusion networks; a high level of online activity has a positive impact on knowledge diffusion; and polarized emotions have little influence on the knowledge diffusion processes. We conclude this paper with a discussion of how exponential random graph models could contribute to our understanding of the structure of online knowledge diffusion networks.