All-Words Domain Specific WSD task
Siva Reddy
In the trail data of SemEval All words domain specific WSD task, I did not find any relation between the background documents and trail data given. Why is this so? Is it the case with final data too. Also background documents contains lot of junk. Does this apply to final data?.
Many tools based on WordNet (like similarity packages) does not accept LMF format. Can the LMF format be converted into standard WordNet dictionary files. Are there any tools available to do such conversion? What kind of tools are available for WordNet-LMF format?
Any ideas?
Thank you,
Siva
Eneko Agirre
Hi Siva,
thanks for your interest!
First of all the train data that was made available was done with the
purpose of showing the data formats, so the actual content is not
related to the target domain.
The test documents will be related to the background texts that we provided.
The background texts have been automatically mined and converted to text
from html and pdf documents. The users of the Kyoto project provided the
url's. The background documents you can download are the result of the
automatic processes, as in a real user scenario. We also provide the
actual urls.
Regarding WordNet LMF, this is the proposed standard representation for
wordnets, and the last releases of wordnets are all following this
representation. The English WordNet was released in an internal format,
which is not available for other language wordnets.
As far as I know there is no tool to convert from any database-oriented
formats to the internal format of the English WordNet. There is the
grinder program released with the English WordNet, but it only works
over the so-called lexicographers files. Perhaps the developpers of
the Simalirty package have more information.
Internally, we have programs to operate over LMF, e.g. a conversion
program from LMF to the dictionary and relation textual files needed by
UKB (another similarity package), which turns out to be easy to browse
and process. If anyone is interested, send me a note and we will happy
to send you the code, but there are no guarantees.
best
eneko
> --
> You received this message because you are subscribed to the Google
> Groups "SemEval-2010 task 17: all-words WSD on a specific domain" group.
> To post to this group, send email to WSD-d...@googlegroups.com.
> To unsubscribe from this group, send email to
> WSD-domain+...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/WSD-domain?hl=en.
--
----------- NEW URL: http://ixa2.si.ehu.es/eneko ------------
Eneko Agirre .
Informatika Fakultatea mailto: e.ag...@ehu.es
Manuel Lardizabal, 1 .
20.018 Donostia fax: (+34) 943 015590
Basque Country (via Spain) tel: (+34) 943 015019