[developers] [incr tsdb()]: reconstructing other MRS types when thinning

Stephan Oepen oe at ifi.uio.no
Wed Mar 9 20:24:33 CET 2011

hi peter,

[incr tsdb()] is in principle agnostic about which MRS serializations  
are used (or, in fact, which meaning representation language more  
generally), but there is a clear bias in the implementation for the  
‘simple’ MRS format.  in my view, this format strikes a good  
balance between human and machine readability and space consumption.   
hence, wherever possible i would recommend ‘simple’ MRSs over XML- 
based alternatives.

to answer your specific question, during (thinning) normalization the  
[incr tsdb()] parameter *redwoods-semantics-hook* can provide a  
function to receive a reconstructed edge and return a string; the  
string will then be recorded as the semantics field.  the predefined  
mrs::get-mrs-string() returns the ‘simple’ format, but defining a  
variant of this function should not be hard.  alas, as things stand  
just now, a user would have to provide an adapted semantics hook  
themselves (and patch the ‘redwoods’ script to activate it).

best, oe

On 9. mars 2011, at 18.04, Peter Adolphs <Peter.Adolphs at dfki.de> wrote:

> Hi!
> Does anyone know how to reconstruct MRS types other than simple  
> format (MRX,
> RMRS, DMRS) when thinning a profile with [incr tsdb()]? Is it  
> possible to
> specify the desired MRS type as a simple parameter?
> Cheers,
> Peter
> -- 
> Peter Adolphs
> DFKI GmbH, Projektbuero Berlin, Alt-Moabit 91c, D-10559 Berlin
> Telefon: 030 / 3949-1826
> _______________________________________________________________
> Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
> Firmensitz: Trippstadter Strasse 122, D-67663 Kaiserslautern
> Geschaeftsfuehrung:
> Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender)
> Dr. Walter Olthoff
> Vorsitzender des Aufsichtsrats:
> Prof. Dr. h.c. Hans A. Aukes
> Amtsgericht Kaiserslautern, HRB 2313

More information about the developers mailing list