#relevance feedback

[betapoc]

Use of clickstream data (what links a user clicks on) to provide indirect relevance feedback for IR.

[rockymedure-blog]

Reading for the Week of 2.27.12

This week we arte learning about relevance feedback and query expansion techniques. It's very interesting to learn about because as a general user of search engines (mainly Google) I've often wondered about the different ways in which they enhance my ability to resolve and information need. 

Relevance feedback is a very interesting concept.  I would have enjoyed trying out the Excite.com feature for user's to provide RF.  However, I suspect that it was less-than-effective b/c when a user has an information need that must be handled quickly, then having the user go a couple of extra steps could be the difference b/w a site getting enough traffic to survive and simply withering on the vine an die.  Furthermore, I wonder what it's like for these search engines, now that billions of queries cross their logs?  What can be learned from the query logs and how do they extract the info?

Query expansion is something that most user's are very familiar with.  We see query expansion as a natural part of a given search session.  I wonder just how many searches are extended in this way?  It certainly would be fascinating to take a peek under the alrgorithmic covers to see what the search giants are implementing in order to make us smarter.

Until next time...

Rock