To take place on the 13th and the morning of the 14th, with small group activities (see FeforSmallGroups) proceeding throughout the week at Fefor. FeforRmrs will take place on the afternoon of the 14th and morning of the 15th. The emphasis in the processing sessions is intended to be on practical activities that are best done face-to-face and where we can achieve concrete results while at Fefor. In particular, getting existing resources working more widely and improving the wiki documentation. We also hope to make progress on ongoing collaborative activities, especially those that require negotiations of interface standards and so on.

Participants should feel free to add additional information to sessions (obviously large scale additions are best done with new pages), but should email AnnCopestake with suggestions of changes to schedule and topics. Names below are of presenters and also of proposers of discussion topics: leaders of discussions and scribes will be added when we can work out load. Please email AnnCopestake to volunteer.

• Automated testing (FrancisBond, UlrichSchaefer, cf. also TestingWithAnt)

  • Everyone agreed it was a good idea, and that we neeed a framework people can add too
  • Ulrich described some of their success with ant, and noone objected to its use

• Action Points

  • Ulrich and Oe: look at getting HoG and PET into the standard distribution tree

  • CJ and Ann: look at building a framework based on ANT for testing with SciBorg at Cambridge

• FeforStandoffAnnotationInterface / SMAF (BenWaldron and UlrichSchaefer)

  • Everyone: collect examples of deep/shallow interface issues so we can design a language for the interface

• Preprocessors (FrancisBond)

• PET API (proposal, discussion) (UlrichSchaefer)

  • Uli to start outline on the wiki: FeforPetApi

• Grammar interfaces: clear interface for robustness-related grammar elements (paths/types/pos-tags/expected tokenization/punctuation etc.) e.g. specified in the inline documentation in such a way that it could be processed/extracted automatically from grammar sources. Initial discussion here: later discussion in plenary session if appropriate.

  • General Issues discussed, more specific issues → working group (led by Ben)
  • Document the proposed lexical type naming (Dan?)

• Presentation/Demonstration: Trollet: tool for multilingual grammar engineering - a wrapper around the LKB (PavelMihaylov)

• Now plenary session only