How to use wordnet:querydata to find out all the related words and terms?

[mmnn]

How to use wordnet:querydata to find out all the related words and
terms?

For example, when I type in "travel", I can get " trip, change
location, journey....." and all of the words and terms related.

Thanks a lot!

[Thanassis Lykartsis]

It's more or less what I was going to ask! Crazy coincidence! :P

Anyway, more formally put, I have a set of terms, which scales up to 10000+, and I want to compute similarities between them. Obviously this adds up to 50 million comparisons, which is not feasible using wn::similarity, since every one comparison needs about 3 seconds on a fast pc, regardless of which similarity measure is selected. My initial approach was to find a way to pick, say, 100 of the terms, and do the trick with them. But then I figured: is there a way, apart from e.g. taking the other members of a synset to which a term belongs, to take a term and then produce all the terms which have > c similarity to this one, with regard to a similarity measure? Sort of like an inverse wn::similarity! :P

Also I would like to ask if there's somewhere around a version of wn::similarity able to handle stems of words.

Thanks!

[Ted Pedersen]

Hi Thanassis,

You might want to move this discussion to the WordNet::Similarity
mailing list - you raise some interesting points, but they are
probably independent of QueryData to some degree.

http://tech.groups.yahoo.com/group/wn-similarity/

However, I don't think you'll find a great answer for your question
about how to give a word and find the X most similar words to it -
that would be great functionality to have, but it's not in
WordNet::Similarity and I'm not sure if anyone else has come up with
that (but it would be worth checking I think).

Cordially,
Ted

--
Ted Pedersen
http://www.d.umn.edu/~tpederse

[Thanassis Lykartsis]

Duely said! Will do...and thanks for the reply!

Also, my congrats on your academical work, I've been through some of your papers and I find it all very interesting...I guess I'll just ask away in wn::similarity! :)