WordNetPlusPlus

[Luciano Del Corro]

Hi Tristan and Nicolai

I have been using DKPro extensivelly, thank you for the effort you put there.

I have a question regarding WNPlusPlus sense inventory. I downloaded the files from the official WN++ webpage (to which you have a pointer) but it seems that the system cannot read those files. It seems that you are expecting the file to contain some confidence for each relation but in the files available in the webpage do not have such a confidence. There is a set of files with synsets and an equivalent set with sense_keys. The format iaccording to the README is --> src_sensekey (<tab>target_sensekey_i)+

Is it possible that you have an acces to a version with confidences? If so can you please point me to the version compatible with DKPro?

Thank you very much in advance
Luciano

[Tristan Miller]

Dear Luciano,

On 23/02/14 12:00 AM, Luciano Del Corro wrote:
> I have been using DKPro extensivelly, thank you for the effort you put
> there.

Thanks! We're glad you find the framework of use. If it leads to
published research results for you we'd be glad for a citation to our
ACL 2013 paper.

> I have a question regarding WNPlusPlus sense inventory. I downloaded the
> files from the official WN++ webpage (to which you have a pointer) but
> it seems that the system cannot read those files. It seems that you are
> expecting the file to contain some confidence for each relation but in
> the files available in the webpage do not have such a confidence. There
> is a set of files with synsets and an equivalent set with sense_keys.
> The format iaccording to the README is --> src_sensekey
> (<tab>target_sensekey_i)+
>
> Is it possible that you have an acces to a version with confidences? If
> so can you please point me to the version compatible with DKPro?

Thanks for spotting this discrepancy. It seems that we coded our
WordNetPlusPlusSenseInventory to read a different file format than the
one from the official WordNet++ distribution. Our program expects a
tab-separated value file with four fields: the similarity score, the
relation type, and the two synsets to be related. Looking through my
files, I see I've got conversions of the official WordNet++ distribution
into the format we support, and will send these to you by private e-mail.

Regards,
Tristan

--
Tristan Miller, Research Scientist
Ubiquitous Knowledge Processing Lab (UKP-TUDA)
Department of Computer Science, Technische Universität Darmstadt
Tel: +49 6151 16 6166 | Web: http://www.ukp.tu-darmstadt.de/

[Luciano Del Corro]

I received it, thanks a lot!!