This one was a lot more work, as there is a lot of code that goes along with it. I'll post a link when I have it running on an externally accessible machine (prob. tomorrow) if anyone wants to try it out.

Its basically a system for refining a web search for a base term (like "patri") when searching for just that term gets you lots of false positives, and you don't want to restrict yourself to a specific auxiliary term ("patri friedman). And you want the computer to figure out good +terms and -terms for you, and learn how to pick out which pages you want. The problem is that it was too big a project and I didn't have time to do it right, I ended up writing the framework and filling in so-so versions of everything. Its just at the stage now where it could start to improve rapidly with some more tweaking. Sigh. Feels so inefficient, it bothers me. Maybe I can find someone who wants a job at Google and will finish it as a resume piece.

<< Funny Flash (6/7/03) << || >> Patri Paper (6/01/03) >>