[developers] Possibly malformed RMRSs from RASP conversion

[Andrew MacKinlay]

Hi,

In the output of the RASP-RMRS converter from the Heart of Gold  
download page: http://heartofgold.dfki.de/pkg/components- 
rasp2.tar.gz , in the output RMRX I'm getting RArgs with labels that  
don't correspond to any known anchor.

For example, in the attached XML (the RASP output that produced it is  
also attached), the following rarg:

   <rarg>
     <rargname>ARG1</rargname>
     <label vid='128' />
     <var sort='x' vid='1' />
   </rarg>

refers back to vid '128', but this doesn't correspond to an EP anchor  
that exists elsewhere in the file.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 1493333.001.rasp.rmrx.xml
Type: text/xml
Size: 18169 bytes
Desc: not available
URL: <http://lists.delph-in.net/archives/developers/attachments/20090807/6c80dd0b/attachment.xml>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: 1493333.001.rasp.deriv.txt
URL: <http://lists.delph-in.net/archives/developers/attachments/20090807/6c80dd0b/attachment.txt>
-------------- next part --------------


The sentence in plain text is:
"The active nuclear form of the NF-kappa B transcription factor  
complex is composed of two DNA binding subunits, NF-kappa B p65 and NF- 
kappa B p50, both of which share extensive N-terminal sequence  
homology with the v-rel oncogene product."

I'm getting at least one of these errors in about 2/3 of the sentences  
which RASP can successfully parse (These are long biomedical sentences).

Has anyone got any thoughts on this? Is this a broken RMRS, or are  
these types of non-existent relationships allowed? Either way, the  
only strategy I can see is to discard the problematic rargs (although  
the latter would give me some theoretical justification for doing it).  
Is it worth trying to build the RMRS converter with a newer LKB tree  
(getting a copy of ACL might be possible although not that easy so if  
anyone has a build lying around they could give me that would be much  
appreciated)? Or am i doing something else strange?

Thanks,
Andy