I've been looking at several companies and products recently, who are all trying to derive context/categorization from content. Editorially it's pretty easy to see that a blog post is about "cars," but programatically it's not so straightforward.
The combination of user generated tags around URLs in general (e.g. technorati with blog posts), tags around bookmarks (e.g. del.icio.us with random URLs), and "tags" derived from keyword search terms on a given blog (e.g. lijit for blog search), are a wealth of categorization material.
It's probably time to collapse all these sources of meta-data about a resource, into more broadly useful services.
