For my Master’s Project, I’m going to be doing some empirical analysis of the weblog data published by Inteliseek. Before I get started though, I need to do some background reading, so that I can at least get a sense of what other research has been done (and I’m taking you with me).
My first stop is Cameron Marlow’s work from MIT. He’s the guy behind Blogdex and has been studying information diffusion in weblogs since around 2001.
Today’s reading is Audience, structure and authority in the weblog community. As you may have guessed by now, I’m writing this mostly as an exercise for my research, a way to collect some notes. I’m not going to give a full summary of the paper, but just hit a few high points that interest me.
Marlow observes that weblogging is a social phenomenon, an assumption that seems to pervade discussions around blogs. This is a convenient observation, as it allows us to use research in Social Network Analysis (SNA) while studying blogs.
One of the interesting insights is that it seems that information flows are controlled by a two-step process. First the information flows up to the opinion leaders, then back down from opinion leaders, meaning that opinion leaders are very important for studying how information flows through social networks.
Opinion leaders can typically be identified by their location in the social network. The common measure, at least in blogs is the number of inlinks the blog has received (in SNA terminology, degree centrality). The other central measures (as described by Freeman 1978), are betweenness and closeness. These measures require a good deal more data and computation in order to calculate and are therefore not commonly used. I wonder if pursuing them would be worthwhile. My intuition says no, but I’m certainly curious.
Marlow’s paper continues to show some statistical analysis of ranking based on blogroll and permalink links. For the top sites, there’s a correlation between the two, but as you move further down the rankings, there seems to be little correlation between number of blogroll links and number of permalink links.
Alright, on to the next paper…