Read Me, Love Me!!

August 9, 2004 11:32 AM

WordNet: "an online lexical reference system whose design is inspired by current psycholinguistic theories of human lexical memory. English nouns, verbs, adjectives and adverbs are organized into synonym sets, each representing one underlying lexical concept. Different relations link the synonym sets." What does one do with WordNet?
posted by archimago (16 comments total)

wow. lots that i can think of. too much to go on about right now.
thanks for the link!
posted by ethylene at 11:54 AM on August 9, 2004

I was trying to figure ou the same thing the other day. Hopefully, ethylene will return with a list.
posted by strangeleftydoublethink at 12:00 PM on August 9, 2004

filter!
posted by quonsar at 12:34 PM on August 9, 2004

Sense 3
trickle, dribble, filter

Clearly this site should be renamed MetaDribble.
posted by languagehat at 12:48 PM on August 9, 2004

What does one do with WordNet?

Lots of computational linguistics.
posted by callmejay at 1:05 PM on August 9, 2004

... the same thing you use Hyperspace Analogue to Language or Latent Semantic Analysis for... which is the analysis of extremely large volumes of text for 'meaning'. I have no idea why anyone, including the military, would be interested in such a thing.
posted by snarfodox at 1:07 PM on August 9, 2004

3-3.) Under Windows, the "Print" button is missing when I "Find keywords by substring" and "Print current display" is not found under the "File" menu.

The printing functions are not available in the version of WordNet that runs under Windows. The documentation included in the package erroneously implies that printing is available on all platforms.

Bummer.
posted by rushmc at 1:52 PM on August 9, 2004

Wordnet's site used to be much, much cooler. You could give it any two words, and it would quite literally find a route between them. For example:

Cow -> Bullet

Cow IS AN Animal HAS blood CONTAINS iron COMPOSES A bullet.

I assume there's some way to revive this functionality, but I can't find it.
posted by effugas at 1:53 PM on August 9, 2004

effugas: are you sure you weren't thinking of Lexical FreeNet?

I used it to make a search engine front-end.
posted by sonofsamiam at 2:01 PM on August 9, 2004

These seem like great tools for building a slightly more interesting (if not necessarily coherent) chat bot.
posted by cortex at 2:20 PM on August 9, 2004

When people try to communicate the meaning of the message usually falls somewhere between them. Nobody ever totally understands what somebody else means. And for that matter seldom do we know the extent of what we actually mean ourselves when we speak or write. ambiguity rules, know what I mean? This site can help. Thanks
posted by donfactor at 3:29 PM on August 9, 2004

Sonofsamiam,

You're my hero. I've been trying to track down LexFN for years.

Wouldn't happen to have any interest in developing a linguistic toolkit for cryptographic key representation, now would ya? :-)
posted by effugas at 4:07 PM on August 9, 2004

Surely this could be of some use in machine translation? Most of the errors I see in translations seem to be with it thinking a word is being used in a different sense to how it's actually being used.
posted by reklaw at 4:51 PM on August 9, 2004

... nor is it likely that anybody would be filtering large quantities of data before giving flagged content to human analysts for further in-depth examination.
posted by snarfodox at 1:17 AM on August 10, 2004

It's possible that Google has been using some type of latent semantic indexing to display text ads in the right context on pages that display those google ads.

It's also possible that Google and other search engines, are, or will be, using something similar in their indexing of pages on the web.
posted by bragadocchio at 3:58 AM on August 10, 2004

bragadocchio: I'm not sure exactly which methods are deployed, but there are certainly plenty of people in the research community using semantic representations in information retrieval. My personal opinion is some such representation is essential, although I generally dislike the use of knowledge base systems like WordNet - I'd prefer to see a system which could learn a low dimensional representation from data.

In some sense, pseudorelevance feedback falls into that category - an idea which has been around for years and is very widely used.
posted by Singular at 4:21 AM on August 10, 2004