Tuesday, September 21, 2004

Guess what

I found out tonight that when my professor was told about my text mining group's project idea, he said "If this works we're taking it to the media." Not sure if that means this is an exciting project or an impossible project.

7 comments:

Anonymous said...

Nobody has explained to me yet what text mining is.

Anyway, what are you mining for? Why would it be exciting/impossible?

Is it Chinese?

serapio said...

Text mining is a kind of magic where you take a big pile of text, mush it around, and then squeeze out something that wasn't there in any of the texts when you started. Like if you had a big pile of texts that each compared a few ways of skinning cats, and you make your program construct a taxonomy of cat taxidermy from the collection. No cats were injured in the production of this illustration.

Our project is to assign bias ratings (and maybe labels) to web pages. Either from the network, the way biased pages tend to refer to each other whereas more neutral pages tend to refer to a variety of pages, or from the language itself, the way biased and neutral pages differ in their usage, or from a combination of the two.

slowlane said...

A bias rating? Like pointing out that one webpage is only slightly biased and another one is EXTREMELY biased? That would be really cool. It would make getting balanced information online way easier.

caedmonstia said...

I dare you to find any webpage that is neutral. But, like our fellow commentator mentioned, perhaps it is possible to differentiate the biased from the EXTREMELY biased. I will create a webpage about Bush which you will find helpful to start off with.

serapio said...

That's quite true. Everybody has a bias. The point is to rate how strongly it shows, or how one-sided their discussion is.

And that one about Bush would be quite helpful. I'd like to order one of those and about 499 others like it about other topics. And 500 that are approximately neutral.

slowlane said...

Ooo. That would be so fun. You should have a contest where you have people send in weird biased urls. I know of several people who would have tons of fun pointing them out to you.

the girl in the red hat said...

It was such a nice idea and so much fun while it lasted