R: Open Issues in the Model

Dear John, Philipp, all

 

I’m in greece for a project meeting. Here a few quick comments, then, yes, I’ll be back Thursday night, so available on Friday for the telco.

Metadata

* There is no link between the metadata module and any other module in OntoLex

Yes u r right. Though easy to add when finished.

* The Lexicon class is a duplicate of one already in the core

Well, actually, the contrary ;-) we originally proposed the class in the metadata module, when depicting the triad: lexicon, lexicalization, ref onto

I’m in no way criticizing the introduction of “Lexicon” even in the core, though, let me recap just the warnings I already gave at the time of that discussion:

 

1)      A lot of vocabularies are actually thought to depict datasets which have a kind of identity, though, this “identity” (intended as the collection of , or better an umbrella over, some kind of data) is usually not explicitly represented in the data. The sole (notable) exception is owl:Ontology (also, owl is actually a metavocabulary as it is used to design vocabolaries which are in turn used for representing data). This is one point against “Lexicon” in the core.

2)      Metadata vocabularies are following a kind of indirection provided by proxy objects such as void:Dataset. This is one point “in favor” of putting the Lexicon in the metadata, as a proxy for the Lexicon data depicted through the core vocabularty (and other data related modules)

3)      Solutions such as voaf:Vocabulary (the proxy for owl:Ontology) suggests to keep the Lexicon in the metadata even if it is also in the core.

* The language property is a duplicate of one defined in the core

Yes, alread discussed. To be updated.

* ConceptualizedLinguisticResource is not used by any other part of module

This is part of the old Lime. We agreed on removing it, as it seems we were not too much for the descriptioin of (various kinds of) LRs. Then, the problem (think, to your disappoint :D ) emerged again as there have been more than one proposal for getting back on the representsation of (even poorer, such as entry:set-of-translations) lexical resources. Despite the horrible name (which in any case had to be changed), I decided to keep the class until we came out with a defoinitive decision on the lex res thing.

 

* The 'lexical link set' class and property are not used by any other part of the module

Yes, this was a more recent proposal for labeling somehow linksets between ontologies and LexicalConcepts of lexical resources (no lex entries playing any role). This was left out from the last discussion on LIME.

 

* Several properties are named the same as classes except for the case of the first letter: resourceCoverage, language, lexicalLinkSet and lexicalization

Don’t recall, did we agree to avoid this pattern? If yes, no problem, we can rename them. Personally, I prefer this pattern but, no problem if it was agreed to avoid it.

 

* Is the 'linguistic model' really required by every lexicalization?

Erm…would say yes :) 

 

Cheers,

 

Armando