Ontolex Resource View: Lexicalization

[Roland Wingerter]

Dear VB3 team,

in all my Ontolex projects, the lexicalization partition of concepts is empty. I would expect to see a concept's lexicalizations, but it is empty.

Am I missing something?

Kind regards

Roland

[Manuel Fiorelli]

Hi Roland,

in OntoLex projects, the section lexicalizations in the resview (of ANY kind of resource) collects denotations given as a plain triples using the predicate ontolex:isDenotedBy. The OntoLex-Lemon specification also defines its inverse ontolex:denotes and the possibility of reify a denotation through an ontolex:LexicalSense, which is related to both a lexical entry and domain resource through properties each provided with an inverse.

By default,  VocBench instantiates all different patterns for giving a lexicalization, so you should always have triples with the property ontolex:isDenotedBy.

Conversely, existing datasets imported into VB might provide denotations only in reified form (through sense resources) or just in the description of the lexical entry that denotes a given resource. In these cases, the description of a "concept" contains no triples with predicate ontolex:isDenotedBy, and thus the lexicalizations section is empty. You can solve this problem by generating plain denotations for the resources using a simple SPARQL update. Perhaps, GraphDB may be capable of inferring them using a suitable ruleset (i.e. one for OWL that includes property chain axioms).

If by "concepts" you meant "lexical concepts" (i.e. instances of the class ontolex:LexicalConcept, which is a subclass of skos:Concept), the problem is different: the binding of lexical entries to concepts (called conceptualization) is represented using a dedicated set of properties, in particular ontolex:evokes and ontolex:isEvokedBy for the plain representation. In this cases, the lexicalizations section will always be empty. Probably, we should improve this aspect, by a dedicated section or extending the definition of the lexicalizations section.

Regards,

Manuel

--
You received this message because you are subscribed to the Google Groups "vocbench-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to vocbench-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/vocbench-user/9eda0f71-5e04-4a1e-9b6e-7eae190406efn%40googlegroups.com.

--

Manuel Fiorelli

[Roland Wingerter]

Hi Manuel,

indeed I meant ontolex:LexicalConcepts. You can see what I mean when you install OMW (pwn-concepts.rdf and wn_08_eng.rdf) as described on the VocBench website). It uses ontolex:evokes/ontolex:isEvokedBy to connect lexical entries to lexical concepts, as foreseen in the Ontolex model.

The problem shows when a lexical concept is evoked by many lexical entries, e.g. 00941990-v "talk (en), speak (en), mouth (en), utter (en), verbalise (en), verbalize (en)". In a case like that I really miss the lexical entry's written representations in the resource view's Lexicalization section. The tab header simply does not have enough room to accommodate all of them. The missing lexicalizations might not seem a big deal as long as the current concept is also displayed in the structure view on the left, but when you navigate in the resource view from one concept to the other, this is no longer the case.

Probably, we should improve this aspect, by a dedicated section or extending the definition of the lexicalizations section.

Thank you, this improvement would be very welcome.

Kind regards

Roland

[Roland Wingerter]

Hi Manuel,

here is a specific use case. I have a VB3 project which aims at creating a German Wordnet by linking de_dbnary (http://kaiko.getalp.org) to OMW (pwn-concepts.rdf and wn_08_eng.rdf).

In this Ontolex project, there are two different paths from a lexical concept to a lexical entry:  

1. LexicalEntry evokes/isEvokedBy LexicalConcept .

2. LexicalEntry sense/isSenseOf LexicalSense .
LexicalSense isLexicalizedSenseOf LexicalConcept .

OMW uses evokes/isEvokedBy while dbnary has an explicit lexical sense in between.

I include two attachments, one with a lexical concept and English lexical entries (PWN), the other one with lexical entries in German (de_dbnary).
Here is a specific use case. I have a VB3 project which aims at creating a German Wordnet by linking de_dbnary (http://kaiko.getalp.org) to OMW (pwn-concepts.rdf and wn_08_eng.rdf).

In this Ontolex project, there are two different paths from a lexical concept to a lexical entry, both specified in the w3c Ontolex Recommendation

1. LexicalEntry evokes/isEvokedBy LexicalConcept

2. LexicalEntry sense/isSenseOf LexicalSense
LexcicalSense isLexicalizedSenseOf LexicalConcept
OMW uses evokes/isEvokedBy, while dbnary has an explicit lexical sense in between (see attachments with sample data for "woman, adult female" from PWN and dbnary). It would be great if VB3 would support both ways to connect lexical concepts and lexical entries.

Kind regards
Roland

[Roland Wingerter]

Sorry, my previous message contains duplicate text (copied and pasted). - Roland