Version 1 (modified by 9 years ago) (diff) | ,
---|
=Notes from FCS-taskforce workshop meeting 2015-05-29=
What info can we provide to foster the community to work together collaboration
- wiki template
- software stack
==Advanced FCS== *Goal: Spec in 2-3 months, endpoints end of year
- Agree on features
- Other annotation layers.
Productive dicsussions. Start with low hanging fruit.
What are essential decisions in order to be able to move forward?
==POS== Different tagsets. Tools or language? Universal dependencies. Mappings.
==MSD== There is already a working group in Clarin-D for this to find use cases. First results from the group in a few weeks.
==Layers== As discussed before: fulltext, lemma, token/word, pos, orth trans, phon trans, headword, ner, gestures, signs, discourse, translation, depency trees, constituents, glosses
QL: cql -> cqp, subset
type a word results/not results boolean
Easy downgrade in case of failure, not hits if no resources are found.
==Communicate capabilities== Features, max better than dynamical minimal set?
==Output/result transport format for advanced-FCS== Focus on overview level today and save details for later
- Multi-layered approach
Does it sound and look like Folia? Use TCF, TEI? Make comparison with real examples.
Bart stand-off will be a problem. Aggregator has to do something with it?
==Summary Dataviews==
- We should keep generic dataviews.
- Make comparison of current or inhouse transport format.
- Yet again we need real world examples!
- Go back to home institutions and we give it some weeks then we have a videoconference with real examples.