In the academic field of information retrieval, this is called "relevance feedback."
Indeed. And if you want to experiment with relevance feedback, take a look at the Xapian Project for a highly scalable
GPLed implementation of the Probabilistic Information Retrieval Model (which is derived from Bayes Theorem, the basis of all those
Bayesian spam filters).
The only search demo up at the moment is the
search over the site itself, which doesn't
show off the relevance feedback especially
well - the pages on the site cover a rather
narrow topic area, which is fine for searching
the site, but less good as a demo. But if
you try it you'll see it suggests various
related words and you can click on these to
add them to the search. Additionally you can
click a checkbox next to a hit to indicate that
it's relevant - check a few and hit search and
the suggested words will be based on those documents you liked.
Slashdot Mirror
Indeed. And if you want to experiment with relevance feedback, take a look at the Xapian Project for a highly scalable GPLed implementation of the Probabilistic Information Retrieval Model (which is derived from Bayes Theorem, the basis of all those Bayesian spam filters).
The only search demo up at the moment is the search over the site itself, which doesn't show off the relevance feedback especially well - the pages on the site cover a rather narrow topic area, which is fine for searching the site, but less good as a demo. But if you try it you'll see it suggests various related words and you can click on these to add them to the search. Additionally you can click a checkbox next to a hit to indicate that it's relevant - check a few and hit search and the suggested words will be based on those documents you liked.