30 likes | 132 Views
Conclusions – Observations (maybe biased) . Field linguistics: Re-doing the path we did, asking the same questions, finding the same solutions , etc. Which means we have not been able to spread out knowledge enough, if the same things had to be re-invented by a different community!
E N D
Conclusions – Observations (maybe biased) • Field linguistics: Re-doing the path we did, asking the same questions, finding the same solutions, etc. Which means we have not been able to spread out knowledge enough, if the same things had to be re-invented by a different community! • Another proof of the need of an action like CLARIN towards SSH, BUT with a clear involvement of major figures of SSH ... Otherwise it remains too abstract, “detached”. • Different requirements come from field linguistics that may not be covered by what is done for NLP: this has to be checked. If so (for sure yes) important to re-check the model. • Speaking about LMF, it is important to have some exemplary field linguist LDB in LMF & advertise its usage. • No mention here of all the aspects that are agreed, like modularity, 8interoperability for sharing, ... • No standard is completely good for anyone! This should be clear, because by definition a standard is a compromise. Nijmegen, August 2010
Conclusions – Observations • Someone said: No need of standards, just my format. We should make it clear how to use the standards, not necessarily as your own working format, but to share data with the community. Important mappability. What others did may also help avoiding mistakes or speeding up your work. • Concepticon: brings up again the need of analysis of the relation between the Lexicon and the Ontology. We had a preliminary meeting last year, we could have another, e.g. in Berlin? • TEI is good for serializing LMF? Proposal of a joint ISO-TEI activity, BUT careful at which message we give to NLP (of going back?). • Ontology of DCs. Useful, but is it more easily acceptable? Or just move the problem to a different level, and even add to complexity, because you have to agree also on relations. We should check GOLD & EAGLES-ISLE DCs • Problem of communication between different communities: a big problem. Interfaces should be good for the other community not for ours. Repurpose content by developing multiple presentation views. This is also an issue that CLARIN should reflect on. • Surprised that no one mentioned using images.,.... but Nijmegen, August 2010
Conclusions – Sustainability of standards: • Need a life-cycle management for standards. Also standards should be maintained. Not easy, no funding, etc. Keep a community alive. • LMF used a lot! MPI plaid a great role, also with LEXUS. Exciting, BUT no explicit mapping, different interpretations, … • Different types of lexicons. A continuum. We need to provide easy to use examples. Need to instantiate the abstract meta-model in specific data models with specific DC selections for major lexicon types and for different lexicon types. Validate them? Make the examples public & publicise. So that different communities have them ready to use, if they want. This has to be done. • Pay attention to the process of standardization of DCs, to avoid bad mistakes that can be bad for the image with the community. See yesterday wrt GOLD. Importance of doing the right things but also of giving the right messages! • Importance of tools. These can be useful also to enforce same instantiations of the standards. But ... • State general methodological criteria. • Strategic vision: Create a community-based framework for standard use, maintenance, access, ... • Train young people, but first make them interested! But just heard that Standards are not attractive!! Nijmegen, August 2010