- From: Flavia Sciolette <flavia.sciolette@ilc.cnr.it>
- Date: Fri, 14 Apr 2023 16:57:17 +0200
- To: Gilles Sérasset <Gilles.Serasset@univ-grenoble-alpes.fr>
- Cc: Francesco Mambrini <f.mambrini@gmail.com>, public-ontolex@w3.org, Christian Chiarcos <christian.chiarcos@gmail.com>
Thank you all for your answers and suggestions! In this first phase of the work, I think I will follow the advice of those who suggested directly instantiating resource values as individuals. At the same time I will post an issue on git to suggest integrations (I noticed that in Olia the diminutive and endearment seem united and I am thinking about what other examples to bring, besides Italian). I take the opportunity to ask for another opinion. With this formalization: <http://example.org/mylexicon/politicastro <http://example.org/mylexicon/politicastro>> to ontolex:Word; rdfs:label "politicastro"@it; lexinfo:partOfSpeech lexinfo:noun; olia:hasEvaluativeFeature myfeatures:pejorative. one could have a triple inference like this: <http://example.org/mylexicon/politicastro> rdf:type olia:LinguisticAnnotation . (because hasFeature has the domain "linguisticAnnotation"). Could this give reasoning problems in your opinion? Thank you in advance, all bests. Flavia Gilles Sérasset <Gilles.Serasset@univ-grenoble-alpes.fr> ha scritto: > Dear Flavia, > > In DBnary, some languages have morphology modelled extensively (we > do not model the word formation processes). > > For this, I use either lexinfo or Olia depending on the language. > E.g. for German, I have to use OliA because I was missing some > classes and relations. For French, lexinfo is sufficient but I’m > considering using OliA to have interoperability between languages. > > Regards, > > Gilles, > >> On 12 Apr 2023, at 13:14, Christian Chiarcos >> <christian.chiarcos@gmail.com> wrote: >> >> Dear Flavia, dear all, >> >> as for the different options: >> >> (1) If you look for a quick solution, best create your own >> inventory and map it to LexInfo features (using owl:sameAs links), >> or to LexInfo or OLiA classes (using >> rdf:type/rdfs:subClassOf/owl:equivalentClass links), as suggested >> by Alik. This is the fastest approach, and it's also the cleanest, >> because if you're incorrect in your equations, these can be >> rectified without touching your data, but only your vocabulary. You >> can even postpone the linking. (You should then distribute your >> vocabulary along with your data, still, or, alternatively, deposit >> it under a resolvable URL. If you don't want to provide a proper >> ontology, please provide a list of URIs for >> terms/tags/abbreviations you used so that others can help you later >> on to come up with an ontology.) >> >> (2) If you look for a solution with maximum interoperability with >> other OntoLex resources, try to use LexInfo directly. There might >> be gaps, but as Fahad pointed out, you can suggest LexInfo >> extensions. The downside is that it might take a while until they >> are processed -- and maybe they won't be approved at all. I would >> encourage you to suggest to extend LexInfo with pejorative and >> augmentative (either as class or as feature value) because both are >> relatively widely needed. As for endearment, I am less certain, >> could you demonstrate that there are multiple languages >> (preferrably not just dialects of a single language) where this >> category is morphologically different from diminuitive? For help >> with LexInfo, this mailing list is the right point of contact ;) >> >> (3) You can also use your local vocabulary and LexInfo together. >> So, if there are apparent gaps in LexInfo, nothing gets lost, even >> if LexInfo isn't extended (yet). >> >> (4) If you look for a solution with maxiumum interoperability with >> corpora and/or annotation tools, you can use OLiA. However, >> normally, OLiA isn't used directly, but instead, you should create >> your own vocabulary ("annotation model") and then link it with OLiA >> classes (as suggested by Alik). Note that LexInfo is an OLiA >> annotation model, as well. So, this option is effectively the same >> as the first. Like option (2), OLiA can be extended with GitHub >> issues and pull requests. If you need help with that, you can also >> mail me directly. >> >> (5) Instead of OLiA, you can also use the GOLD ontology (the >> terminology is a bit different, but you also should create and link >> your own vocabulary/"community of practice extension", then). I >> wouldn't advise to do so at the moment, for two reasons: On the one >> hand, GOLD hasn't been properly maintained since the 2010 release, >> and it is thus not extensible; on the other hand, GOLD is already >> linked with OLiA, so if you really need a mapping to GOLD, you can >> bootstrap it via options (1)-(4). Same for ISOcat. From OLiA, you >> could also bootstrap a LexInfo mapping, but I would advise against >> it, because using LexInfo directly would be the more natural >> approach for lexical resources. >> >> These features can be attached with lexinfo:morphosyntacticFeature >> (or subproperties) to ontolex:LexicalEntry or ontolex:Form. As for >> treating these features as word formation processes: In addition to >> Francesco's proposal, you can also add these morphosyntactic >> features to either morph:Rule or morph:Morph, either directly with >> "lexinfo:morphosyntacticFeature" or as a bundle of morphosyntactic >> features in the "morph:grammaticalMeaning" object of rules or morphs. >> >> Best, >> Christian >> >> Am Di., 11. Apr. 2023 um 19:43 Uhr schrieb Francesco Mambrini >> <f.mambrini@gmail.com <mailto:f.mambrini@gmail.com>>: >> Dear Flavia, >> Another option might be to explicitly model the derivation process >> between lexical entries, using morphemes and derivation rules. >> You may be interested to see how we did this in LiLa, using the new >> Morph Ontolex module: >> https://aclanthology.org/2022.ldl-1.10.pdf >> <https://aclanthology.org/2022.ldl-1.10.pdf> >> https://publires.unicatt.it/it/publications/the-two-approaches-to-word-formation-in-the-lila-knowledge-base-o >> <https://publires.unicatt.it/it/publications/the-two-approaches-to-word-formation-in-the-lila-knowledge-base-o> >> >> >> Best, >> Francesco >> >> >>> Il giorno 11 apr 2023, alle ore 19:05, Alik Kirillovich >>> <alik.kirillovich@gmail.com <mailto:alik.kirillovich@gmail.com>> >>> ha scritto: >>> >>> Dear Flavia, >>> >>> You can also create your resource-specific features, but define >>> them as instances of the popular OLiA [1] or GOLD [2] ontologies. >>> (OLiA is more suitable to model corpus annotation schemas, but can >>> be used in lexical resources too). In this case, your resource >>> retains interoperability with other OLiA/GOLD-based resources and >>> applications. >>> >>> For example, define the pejorative feature: >>> >>> @prefix : <http://example.org/myfeatures# >>> <http://example.org/myfeatures#>> . >>> @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns# >>> <http://www.w3.org/1999/02/22-rdf-syntax-ns#>> .. >>> @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema# >>> <http://www.w3.org/2000/01/rdf-schema#>> .. >>> @prefix owl: <http://www.w3.org/2002/07/owl# >>> <http://www.w3.org/2002/07/owl#>> . >>> @prefix olia <http://purl.org/olia/olia.owl# >>> <http://purl.org/olia/olia.owl#>> .. >>> >>> :pejorative >>> a olia:PejorativeEvaluative; >>> rdfs:label >>> "pejorative"@en, >>> "peggiorativo"@en. >>> >>> And then link it to the personacce lexical entry: >>> >>> @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns# >>> <http://www.w3.org/1999/02/22-rdf-syntax-ns#>> .. >>> @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema# >>> <http://www.w3.org/2000/01/rdf-schema#>> .. >>> @prefix ontolex: <http://www.w3.org/ns/lemon/ontolex# >>> <http://www.w3.org/ns/lemon/ontolex#>> .. >>> @prefix lexinfo: <http://www.lexinfo.net/ontology/3.0/lexinfo# >>> <http://www.lexinfo.net/ontology/3.0/lexinfo#>> .. >>> @prefix olia <http://purl.org/olia/olia.owl# >>> <http://purl.org/olia/olia.owl#>> . >>> @prefix myfeatures: <http://example.org/ >>> <http://example.org/>myfeatures#> . >>> >>> <http://example.org/mylexicon/politicastro >>> <http://example.org/mylexicon/politicastro>> >>> a ontolex:Word; >>> rdfs:label "politicastro"@it; >>> lexinfo:partOfSpeech lexinfo:noun; >>> olia:hasEvaluativeFeature myfeatures:pejorative. >>> >>> If you use OLiA, the easiest way to link features to lexical >>> entries/forms is to reuse subproperties of the >>> olia_system:hasFeature object property (e.g. >>> olia:hasEvaluativeFeature as in the example below). However, >>> defining your own object properties (as subproperties of >>> lexinfo:morphosyntacticProperty) may be a more preferable way. >>> First, the domain of olia_system:hasFeature is the >>> olia_system:LinguisticAnnotation class, and so, strictly speaking, >>> this property is a property of annotations (i.e. token-level >>> entities), not lexical entries or forms. And, second, such >>> subproperties are not defined for all feature classes. Also, you >>> can link all the features just by the single top-level >>> lexinfo:morphosyntacticProperty property. >>> >>> Best, >>> Alexander Kirillovich >>> >>> [1] https://acoli-repo.github.io/olia/ <https://acoli-repo.github.io/olia/> >>> >>> [2] >>> http://web.archive.org/web/20131118135345/http://linguistics-ontology.org/gold-2010.owl >>> <http://web.archive.org/web/20131118135345/http://linguistics-ontology.org/gold-2010.owl> >>> вт, 11 апр. 2023 г. в 19:09, Fahad Khan <fahad.khan@ilc.cnr.it >>> <mailto:fahad.khan@ilc.cnr.it>>: >>> Dear Flavia, >>> DiminuitiveNoun exists as a class in lexinfo. The others don't, >>> but you can open a new issue >>> <https://github.com/ontolex/lexinfo/issues> on the lexinfo github >>> suggesting them as additions to the ontology. >>> Cheers >>> Fahad >>> >>> Il giorno mar 11 apr 2023 alle ore 16:58 Flavia Sciolette >>> <flavia.sciolette@ilc.cnr.it <mailto:flavia.sciolette@ilc.cnr.it>> >>> ha scritto: >>> Dear all, >>> >>> I write this email to ask for your advice. I am working on a >>> conversion of a lexical resource for Italian in an ontolex-compliant >>> format. This starting resource describes lemmas and associated forms, >>> some of which are derived with suffixes (e.g. for the diminutive or >>> the pejorative). This kind of form is described with a feature and >>> associated values: “dim” for diminutive, “end” for endearment, “pej” >>> for pejorative, and “aug” for augmentative. I am having some trouble >>> with the conversion of these values to define forms. I set out to use >>> Lexinfo, but, to the best of my knowledge, I have not found exact >>> equivalents. I thought about using a sub-property of morphosyntactic >>> properties but first wanted to figure out if there was a way to use a >>> popular vocabulary like Lexinfo, rather than maintaining >>> resource-specific values. I apologize for this possibly naive concern. >>> >>> Thank you in advance. >>> >>> Flavia Sciolette >>> >>> >>> >>> >>> >>
Received on Friday, 14 April 2023 15:00:32 UTC