One of the papers presented at the WWW 2007 involves the efforts of researchers at the University of Toronto to provide some interesting ways to search blogs – BlogScope: Spatio-temporal Analysis of the Blogosphere

The application described is online in “preview” form – BlogScope. It’s presently tracking 9.90 million blogs with 76.78 million posts from Blogspot.

In addition to searching these Blogspot posts, it provides some interesting ways of looking at information in its index that we really aren’t seeing from any other blog search engine:

Popularity Curve – provides the ability to visualize the popularity of query terms as a function of time

Burst Detection – identifies and marks bursts–interesting events –on popularity curve in red.

Information Analysis – enables users to zoom in various parts of popularity curve, and can be restricted to a time interval, which lets you view bursts and evolution of various topics.

Correlation Discovery – shows in search results a list of keywords closely associated with the query terms over the selected time window.

Geo Locator – the search results for a query can be visualized on a world map.

I liked the Geo Search a lot. As they tell us in the paper about what they are trying to do:

The analysis paradigm that BlogScope facilitates is segmented in four steps. BlogScope identifies what is ‘interesting’, when it was ‘interesting’, why it is ‘interesting’, and where it is ‘interesting’.

It would be great to see this expanded to blogs outside of Blogspot. I hope that some of the other folks in the blog search space are paying attention. Some great ideas in here.

An expanded paper from a couple of the inventors – Searching the Blogosphere provides some more insight into efforts being made by the team behind this blog search engine.