Regarding my amateur effort in social network analysis, Valdis Krebs wrote:
Social network analysts consider it a 'stronger tie' if both parties link to each other and AGREE that they link to each other. If A and B have the same confirmed links in a project team they are probably performing a very similar role in the project. You can probably write a spider that finds the 'confirmed ties' between blogs.
Your values of 'overlap' from 0.05 to 0.2 are very low... most social network analysts start paying attention when two nodes have links in common > 0.8.
I agree. The data were too sparse to support much analysis. I really just wanted to suggest the possibilities that arise when we instrument our weblogs to provide structured data.
Jon Schull, meanwhile, has worked out some alternate visualizations . The VPython example is cool. If you already have Python 2.2 installed, it only takes a minute to check it out. The effect feels right, though I'm not sure what it means to rotate through three-factor space.
Former URL: http://weblog.infoworld.com/udell/2002/05/30.html#a270