- From: Hinkelmanns Peter <peter.hinkelmanns@sbg.ac.at>
- Date: Tue, 6 Nov 2018 10:18:47 +0000
- To: 'Christian Chiarcos' <chiarcos@informatik.uni-frankfurt.de>, "'public-ontolex@w3.org'" <public-ontolex@w3.org>
- CC: "christian.chiarcos@web.de" <christian.chiarcos@web.de>
- Message-ID: <e0b84ab31c7a4c20885733549cf7c366@sbg.ac.at>
Dear Christian, thank you very much for your response! Our conceptual system is already a simple SKOS ontology. A single concepts looks like this: <skos:Concept rdf:about="http://www.mhdbdb.sbg.ac.at/concept#23250000"> <skos:inScheme rdf:resource="http://www.mhdbdb.sbg.ac.at/ontologies#ConceptualSystem"/> <skos:prefLabel xml:lang="en">Castle</skos:prefLabel> <skos:prefLabel xml:lang="de">Die Burg</skos:prefLabel> <skos:broader rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23200000"/> <skos:narrower rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23251000"/> <skos:narrower rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23252000"/> </skos:Concept> The not so borderline solution would be the creation of a SKOS concept for each used combination of concepts? So the first sense of hûs would look like this as an individual SKOS:Concept: <skos:Concept rdf:about="http://www.mhdbdb.sbg.ac.at/sense#1234"> <skos:inScheme rdf:resource="http://www.mhdbdb.sbg.ac.at/ontologies#SenseProperties"/> <skos:broader rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23250000"/> <skos:broader rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23309000"/> <skos:broader rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23309020"/> </skos:Concept> And the modified lexicalizedSense should look like: <ontolex:LexicalSense rdf:about="http://www.mhdbdb.sbg.ac.at/sense#4709"> <ontolex:isSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/entry#2933"/> <ontolex:reference rdf:resource="http://www.mhdbdb.sbg.ac.at/sense#1234"/> </ontolex:LexicalSense> Maybe we could actually benefit from those extra-layers of information. At the Moment, there is no linking between entries which share the same combination of concepts as a sense element. With that solution, all equal sense elements would point to the same SKOS concepts, but with the lexicalizedSense layer there is even a place to put information not shared by all entries but only belonging to one individual sense. So it might be acceptable not to use the blank node version preferred by you, even if there aren’t that many entries which share a specific combination of concepts as a sense. And yes, we also need a solution for multiple POS statements. Maybe it would be the cleanest solution to add an individual ontolex:Word for each POS statement and link those entries? It leads to the question, how to define an article in the dictionary. Do entries with multiple use cases aka POS exist? Or is it better to have individual entries? And again: Thank you very much! Best Peter Peter Hinkelmanns MA Wissenschaftlicher Mitarbeiter Mittelhochdeutsche Begriffsdatenbank (MHDBDB) | IZMF Universität Salzburg | Erzabt-Klotz-Straße 1, A-5020 Salzburg | Tel. +43 662 8044 4339 | e-mail: peter.hinkelmanns@sbg.ac.at<mailto:peter.hinkelmanns@sbg.ac.at> Internet: CV und Fodok<http://mhdbdb.sbg.ac.at:8300/wordpress/hinkelmanns/> | mhdbdb.sbg.ac.at<http://mhdbdb.sbg.ac.at/> | facebook.com/mhdbdb<https://www.facebook.com/mhdbdb> | twitter.com/MHD_BDB<https://twitter.com/MHD_BDB> Please consider the environment before printing this email Von: Christian Chiarcos [mailto:chiarcos@informatik.uni-frankfurt.de] Gesendet: Dienstag, 06. November 2018 08:17 An: 'public-ontolex@w3.org'; Hinkelmanns Peter Cc: christian.chiarcos@web.de Betreff: Re: MHDBDB: Connecting Senses with multiple Concepts Dear Peter, great to see the MHDBDB being converted to RDF. For your use case, I would actually not model the conceptual system itself in lemon, but rather use SKOS and provide only lexicalization information in lemon. In other words, your concept hierarchy is the ONtology element in lemON and ONtoLex, and this is actually the original use case of lemon. The relation for linking with an ontology is ontolex:reference. The spec requires that "The lexical sense has a single lexical entry and a single reference in the ontology." (As in other cases, I think this is too restrictive.) So, your problem remains the same. The "proper" solution would be to create "lexicalized concepts" in your thesaurus (which can have skos:broader relations to more than one skos concept), and to connect these with your lexical entries via ontolex:reference. A shortcut, if you will, would be to use blank nodes as part of the ontolex linking: mhdbdbLex:sense4709 ontolex:reference [ skos:broader mhdbdbThes:concept23250000, mhdbdbThes:concept23309000, mhdbdbThes:concept23309020 ] . Technically, this would be valid, but it is a bit borderline, because the blank node is not strictly speaking *in* the ontology/thesaurus. (It is certainly not worse than not having an ontology, though, as in most lexicography use cases.) From a modelling perspective, I would prefer the second solution despite these difficulties, as otherwise, the thesaurus grows infinitely with every novel word being added. Blank nodes created as part of the linking are easy to filter out if you need to work with the categories alone. Also, imagine a use case where you have different lexicalizations/lemon models to be linked to a single ontology, and those lemon models may require fusing concepts in a different way. Using blank nodes as part of the linking makes sure that lexicalization- (lexicon-) specific concept fusion can be clearly separated from the (language-independent) concept hierarchy. We have similar difficulties (and no better solution) for multiple parts of speech for the same lexical entry, etc. Best, Christian Am .11.2018, 12:16 Uhr, schrieb Hinkelmanns Peter <peter.hinkelmanns@sbg.ac.at>: Dear Ontolex-List, I’m writing as a team member of the „Middle High German Conceptual Database“ (MHDBDB, http://www.mhdbdb.sbg.ac.at<http://www.mhdbdb.sbg.ac.at/>), an onomasiological dictionary of middle high German. We are planning to convert our 40.000+ dictionary articles to Ontolex. An article consists of one or many senses which refer to one or many categories of our conceptual system: [cid:image001.jpg@01D475BB.C37C3BA0] In the example above you can see that the entry ‘hûs’ has two senses which each point to different categories in our conceptual system. The conceptual system has already been converted to SKOS. My attempt to express the entry ‘hûs’ with ontolex looks like this: <rdf….> <ontolex:Word rdf:about="http://www.mhdbdb.sbg.ac.at/entry#2933"> <ontolex:lexicalForm> <ontolex:Form> <ontolex:writtenRep>hûs</ontolex:writtenRep> </ontolex:Form> </ontolex:lexicalForm> <lexinfo:partOfSpeech rdf:resource="http://www.mhdbdb.sbg.ac.at/grammar#NOM"/> <ontolex:sense rdf:resource="http://www.mhdbdb.sbg.ac.at/sense#4709"/> <ontolex:sense rdf:resource="http://www.mhdbdb.sbg.ac.at/sense#28403"/> </ontolex:Word> <ontolex:LexicalSense rdf:about="http://www.mhdbdb.sbg.ac.at/sense#4709"> <ontolex:isSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/entry#2933"/> <ontolex:isLexicalizedSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23250000"/> <ontolex:isLexicalizedSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23309000"/> <ontolex:isLexicalizedSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23309020"/> </ontolex:LexicalSense> <ontolex:LexicalSense rdf:about="http://www.mhdbdb.sbg.ac.at/sense#28403"> <ontolex:isSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/entry#2933"/> <ontolex:isLexicalizedSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23209000"/> <ontolex:isLexicalizedSenseOf rdf:resource="http://www.mhdbdb.sbg.ac.at/concept#23309020"/> </ontolex:LexicalSense> </rdf> As you can see above, I’m doing something probably not intended by Ontolex: A lexical sense consists of multiple isLexicalizedSenseOf references, mirroring the MHDBDB datastructure. How could our article structure be transferred to Ontolex correctly according to the scheme? Many thanks in advance and best regards Peter Peter Hinkelmanns MA Wissenschaftlicher Mitarbeiter Mittelhochdeutsche Begriffsdatenbank (MHDBDB) | IZMF Universität Salzburg | Erzabt-Klotz-Straße 1, A-5020 Salzburg | Tel. +43 662 8044 4339 | e-mail: peter.hinkelmanns@sbg.ac.at<mailto:peter.hinkelmanns@sbg.ac.at> Internet: CV und Fodok<http://mhdbdb.sbg.ac.at:8300/wordpress/hinkelmanns/> | mhdbdb.sbg.ac.at<http://mhdbdb.sbg.ac.at/> | facebook.com/mhdbdb<https://www.facebook.com/mhdbdb> | twitter.com/MHD_BDB<https://twitter.com/MHD_BDB> Please consider the environment before printing this email -- Prof. Dr. Christian Chiarcos Applied Computational Linguistics Johann Wolfgang Goethe Universität Frankfurt a. M. 60054 Frankfurt am Main, Germany office: Robert-Mayer-Str. 10, #401b mail: chiarcos@informatik.uni-frankfurt.de web: http://acoli.cs.uni-frankfurt.de tel: +49-(0)69-798-22463 fax: +49-(0)69-798-28931
Attachments
- image/jpeg attachment: image001.jpg
Received on Tuesday, 6 November 2018 13:24:27 UTC