StatusMeeting3September2002

• Date: 3. September 2002
• Time: 10:00 to 12:30 am plus 13:00 to 15:00 pm
• Attendees: WolfgangMenzel, KilianAFoth, TomasBy, MichaelDaum
• Topic: towards partial parsing results for the DFG report

Priority list and job assignments

1. * HeiseGrammar: extract chunk annotations for a fake chunker; write a fake chunker for the WCDG (MichaelDaum)
2. * HeiseGrammar: add morphology patterns for german, initially for the nouns; extend the lexicon template system in the WCDG to allow a hierarchical order of templates (KilianAFoth)
3. Extract subcategorial information from the GermaNet; get/write a baseform normalization in oder to access the GermaNet
4. * Experiments with partial parsing:
   • remove constraints; use only a subset of labels (e.g. not marking the first complement) <- is that sensful?
   • weaken top-node constraints; write top-node-constraints of different classes, e.g. a noun is a better top-node than a determiner (TomasBy)
5. NegraCorpus: extend the HeiseGrammar to cover the lexicon of the NegraCorpus
6. * Experiments with chunk info (maybe get a real chunker attached), including writing constraints for the chunker (KilianAFoth)
7. * NegraGrammar: convert the NegraCorpus to dependency annotations; write a memo documenting that (MichaelDaum)
8. NegraCorpus: extract chunk annotations (MichaelDaum)
9. Invent a new CDG predicate =between(pos1, pos2, condition)= which is no more a local constraint but one relying on the linear wordorder of the sentence