[Corpora-List] Corpora Digest, Vol 40, Issue 27

[Federico Sangati]

Hi,

I've implemented evalC available at http://staff.science.uva.nl/~fsangati/
There is both a command line and a GUI version for it.
Besides standard evalB measure, evalC includes per category recall/precision/fscore, crossing brackets, wrong category (correct span).
The documentation is still rather poor, but i'm planning to improve it.
In case you have any problem with it let me know.

Best,

Federico Sangati

Institute for Logic, Language and Computation
University of Amsterdam
Science Park 904, Amsterdam



On 27 Oct 2010, at 12:00, corpora-request at uib.no wrote:

> Message: 3
> Date: Tue, 26 Oct 2010 12:02:12 +0200
> From: Vera Demberg <vera at coli.uni-saarland.de>
> Subject: [Corpora-List] looking for parser output analysis tool (more
> 	detailed than e.g. parseval)
> To: corpora at uib.no
> 
> Hi,
> can anybody point me to a tool that can output statistics about what is 
> going wrong in a parser, i.e. which labels are wrong most often, and 
> stats about what types of brackets are wrong i.e. attaching under 
> category Y instead of category X etc.
> Thanks a lot,
> Vera
> 
> -- 
> Vera Demberg
> Cluster of Excellence 
> "Multimodal Computing and Interaction"
> Saarland University
> Germany

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20101027/88f0fc84/attachment.htm>
-------------- next part --------------
_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora