I’m summarizing (literally) a new study  that will be presented at the 28th European Conference on Information Retrieval (ECIR’06) 2006 in April written by Gilad Mishne and Maarten de Rijke from the University of Amsterdam.  This came to me care of SmartMobs.

Blog searches are different than general web searches  as people are searching for references  to names of people, brands, companies, and so on and locating blogs by theme.  Blog searchers are also more engaged
in technology and politics than general web searchers.

Search behavior observed is similar to that in general web
search engines: users are typically interested only in the first few results returned, and usually issue a very small number of queries in every session.

Using external resources to categorize the blog queries, the researchers uncovered a  blog searcher profile substantially more concentrated on news (particularly  politics), entertainment, and technology than the average web searcher.  In fact, it may be good for blog search engines to identify and exploit the names of people, brands, companies (and parts of them).

I’m glad research is going on with Blog Search as I don’t think there is a good way of doing semantic analysis of blogs or getting a good list of keywords out of them right now…there’s too much noise in blogs for that.

I have used the AdWords Keyword Planning tool to do semantic analysis in the past and it’s generally OK, but when you use it on blogs I often end up picking up the topics and keywords of the AdSense or BlogAds running on the blog - not always what I want.