UMBC ebiquity
Measuring Influence Using Inlinks

Measuring Influence Using Inlinks

Akshay Java, 1:00pm 17 February 2007

Lately, there has been a lot of discussion about measuring influence using inlinks. A recent post by Matthew Hurst talks about Biz360, a market intelligence company that is using a PageRank like metric to measure influence. Their approach is to measure influence using not just inlinks to a blog but also the rank of the blogs linking to it. Inlinks, PageRank and all its variations work out alright if one is trying to define a global ranking for all blogs. But I think we need to start talking about influence more in terms of communities and readership.

The key question here is not “What is a blog’s influence?“, but “whom does this blog have an influence on and how much?”. This is a more difficult problem. I think the role of communities in influence metrics is quite important. To illustrate this point, here is an example of a small community of political blogs and size of the nodes is proportionate to the inlinks.

political blogs

The interesting thing here is that once we identify the community, inlink counts are quite effective in finding influential nodes. On the other hand, if we just look at the top blogs using PageRank alone it would contain blogs for a mixed bag of topics. For example following are top 3 blogs in the dataset using PageRank:

1. boingboing [Misc/Culture]
2. DailyKos [Politics]
3. Endgadget [Technology]

Thus, from a market research perspective,I think, what would matter most is to measure the influence of a blog on a community. Related: We have a few papers that we are presenting at ICWSM and one of them discusses “Modeling Trust and Influence in the Blogosphere Using Link Polarity”, while another paper talks about “Feeds That Matter: A Study of Bloglines Subscriptions” that uses readership-based measures.

Comments are closed.