- From: Armando Stellato <stellato@info.uniroma2.it>
- Date: Wed, 16 Sep 2015 16:34:47 +0200
- To: 'Juan Antonio Pastor Sánchez' <pastor@um.es>, "'Osma Suominen'" <osma.suominen@helsinki.fi>
- CC: <public-esw-thes@w3.org>
- Message-ID: <DUB408-EAS42925C14FE9EF1D0C693F4BA05B0@phx.gbl>
Dear Juan, well I can speak only for myself, but I think to interpret correctly the thoughts of Osma (I guess his expression “you now want to change that” was in the context of your vocabulary), if I say that no, we were not thinking you were in an effort to change SKOS. My hint was (in the context of a suggestion to the UNESKOS vocabulary which obviously you are free to develop and to use) to the fact that, if some properties were not adopted in SKOS, it could be not a gap, but something done by purpose and, for this reason, it could be worth checking why certain possibilities have been explicitly chosen to be dropped. To concretely provide some rationale, and not just blindly relying on the “sacred words of SKOS”, I would start from your example. Obviously SPARQL may be not available, I agree with you that this might be a scenario. So, let’s check the examples one by one: Quoting you here: “One case: From a iso-thes:ConceptGroup (or micro-thesaurus): How can we reach to the top concepts without using SPARQL and applying the SKOS/ISO-THES properties only? In one word: impossible.” In principle, I would say: if you have a ConceptGroup, and this concept group contains concepts, then by accessing them you can then see which scheme they belong to, and then list the top concepts of this scheme (because SKOS provides an inverse of topConceptOf, and it is not by accident, we’ll talk about this later on in this email..). This is plain SKOS (obviously combined with ISO-Thes in order to deal with iso-thes:ConceptGroup). I still got the doubt that maybe I’m missing something here, but in order to fully understand your example, I reached the description of iso-thes as well: http://pub.tenforce.com/schemas/iso25964/skos-thes/ and when reading this: The skos:inScheme (http://www.w3.org/2004/02/skos/core#inScheme) property should be used to indicate the thesaurus to which an instance of skos:Collection applies (see ISO 25964: isPartOf). it left me a bit puzzled. Well, in case you can help me to get the full path which requires this to be reached, and we can check if it really needs more properties in order to be covered. Now, let’s go to a second example you made: the reachability of a skos:Collection. In the SKOS reference/primer, there are (if I am not missing anything) no examples of skos collections being referenced as being skos:inScheme of some given scheme. But I totally agree with you, this is (again, if I’m not missing anything) possible in principle, and there is no immediate way to do the contrary…reach the collections contained in a skos scheme. So yes, clearly you have a need here. But, as much as we need to be open-minded and think about NON-SPARQLed scenarios, you must also take into consideration what the consequences of your proposed property are in all scenarios. I already made my funny example about the huge description. My example still holds – in an even more catastrophic way! - for an HTML page with RDFa content: would you like to get 30.000 concepts attached to the description of a single concept scheme? One answer could be: ok, this carpet cannot be pulled from too many sides, your vocabulary is suitable for some cases, and less for others. But let’s check if your needs cannot be solved in other ways and stretch this carpet both ways ;-) Your need for covering the linking to skos:collections and these iso-thes:ConceptGroups, more than pushing in my mind the need for the inverse of skos:inScheme, evokes a parallelism with skos:topConceptOf which, as I said before, “not by accident” has an inverse: skos:hasTopConcept. Why this one has an inverse? It’s simple: a scheme can have 20, 30, 50? Top concepts, not 30.000. It makes sense to have a property which, from the scheme, provides those bunch of very important concepts. You can do the same for ConceptGroups and have a property which, instead of being the inverse of skos:inScheme, would be a subproperty of the inverse of skos:inScheme So, having: <cg> a isothes:ConceptGroup <s> a skos:ConceptScheme And defining this (just a sample name): uneskos:hasConceptGroup You can state: <s> uneskos:hasConceptGroup <cg> And also imply by reasoning, the following: <cg> skos:inScheme <s> Without any other friendly-fire such as the undesired explosion of triples of the type: <s> uneskos:contains c1, c2, ….c30.000 Again, just my two cents, but in the effort of helping, certainly not of sinking your proposed vocabulary (and even more important, cover certain needs for your scenarios) Cheers, Armando Da: Juan Antonio Pastor Sánchez [mailto:pastor@um.es] Inviato: mercoledì 16 settembre 2015 14:28 A: Osma Suominen <osma.suominen@helsinki.fi>; stellato@info.uniroma2.it Cc: public-esw-thes@w3.org Oggetto: Re: R: UNESKOS Vocabulary and 2nd SKOS version of UNESCO Thesaurus Dears Armando and Osma Thank you very much for your observations. Osma, sorry, but we don't want to change anything about SKOS. We don't require any SKOS changes. I don't know the reason why you say that. Can you explain me? The UNESKOS vocabulary refers to an vocabulary for and specific case that can be used for others thant find useful our work. Your affirmation that we want to change SKOS is absolutely false. Our work complements SKOS, but doesn't replace it. SKOS should continue as is to remain SIMPLE as you think. I think that define a specific vocabulary that complements SKOS and ISO-THES is not forbidden (as far as I remember) and this restriction is a kind of dogmatism that goes against the very essence of the Semantic Web. And I don't like it. UNESKOS is designed to complement SKOS and ISO-THES. It isn't an alternative. If fact, if you get the SKOS dataset for the UNESCO Thesaurus and ignore all the statements that have any UNESKOS property the SKOS representations is acording to SKOS and ISO-THES. Certainly, the dataset have been tested with Skosify. The question is: Really, it's necessary a vocabulary that define inverses for several SKOS properties? In my opinion, we must not forget that Linked Data remains the Web and I think that the Information Architecture principles for SKOS dataset are applicable (and useful). Please, as we note in the reference document: don't think that a SPARQL Endpoint it's always avalaible. Think in the home page of a web site that no have links to continue the navigation to the inside contents. The lack the properties from the Concept Schemes to the Concept Groups (that are intended for represent micro-thesaurus) is the same. Please take a wider view and don't think only in terms of SPARQL, think in terms of information architecture of the dataset so that it can be reused in other ways to increase interoperability. So, please, think about a scenario in which you don't have an SPARQL Endpoint to query the dataset. An example for this: the HTML version of a vocabulary with RDFa markup that include the RDF statements of the SKOS dataset into the HTML markup. The Concise Bound Description doesn't useful in this case because implies needless and artificial HTML code. One case: From a iso-thes:ConceptGroup (or micro-thesaurus): How can we reach to the top concepts without using SPARQL and applying the SKOS/ISO-THES properties only? In one word: impossible. So this is the meaning of uneskos:hasMainConcept and uneskos:mainConceptOf: Access points to continue (or start) to navigate the hierarchical structure of the thesaurus. Even more: How can we reach from skos:ConceptScheme to iso-thes:ConcepGroup. Impossible, because iso-thes:microThesaurusOf goes on inverse way. So this is the meaning of uneskos:hasMicroThesaurus. The uneskos:memberOf solves the discovery of the Collection/ConceptGroup to which a concept belong without use SPARQL. Repect uneskos:contains. Well, with a depply reading of the reference document you can find: NOTE: It is not necessary the inclusion of uneskos:contains property between the Concept Schemes and the Concepts in the SKOS dataset. In the context of a well designed KOS, it is possible to find a path to any Concept of the KOS starting from the Concept Scheme and the Top Concepts. However, SKOS does not provide any element to discover (i.e.) the Collections of a KOS from the Concept Scheme. The property uneskos:contains covers this need. This is because the first SKOS version of the UNESCO Thesaurus includes statement like this: <Collection> skos:inScheme <Concept_Scheme> And the uneskos:contains property is intended for: <Concept_Scheme> skos:contains <Collection> Best regards, Juan 2015-09-16 13:36 GMT+02:00 Osma Suominen <osma.suominen@helsinki.fi <mailto:osma.suominen@helsinki.fi> >: Hi Juan! My first reaction was the same as Armando's 2nd point, i.e. some of your extensions (namely unesco:contains and unesco:memberOf) are extensions of core SKOS with no inherent relationship to the UNESCO Thesaurus. There are certainly arguments about whether properties should be defined in both directions or not. In this case SKOS has decided to make only one way relationships (at least in these two cases, skos:inScheme and skos:member) and you now want to change that by introducing the inverse properties as well. You can certainly do that for your own thesaurus but I think that sticking to the SKOS properties would be simpler for everyone. In SPARQL and most RDF toolkits it is not very difficult to follow property paths in either direction. -Osma On 16/09/15 11:57, Armando Stellato wrote: Dear Juan, just a couple of notes. 1.Why the word “main” adopted in some property names? I don’t find any rationale for the use of “main” as the concepts do not seem to be “main” at all for the related schemes/collections. 2.Some properties, more than being targeted at a special domain (e.g. UNESCO), seems plain extensions of the core SKOS, aiming at filling some gaps left by it. However in some cases I feel like these gaps were left by purpose. E.g. the uneskos:contains provides an inverse property for skos:inScheme. Think about a 30.000 concepts thesaurus (for which there exists at least a scheme containing all concepts). If you were using the UNESKOS vocabulary, you surely would try a SPARQL DESCRIBE on this main scheme? ;-) a.Use of SPARQL (or -1 expressions in DL) does not strictly require for the presence of named inverse properties for everything. So, why do they exist? Because when you get the description (for example by a DESCRIBE, but not limited to that) of a resource, they can provide nice “resumes” for it. The kind of path you want to realize (see the description of hasMainConcept) can be performed without the strict need of the property. It is up to SKOS browsing tools to allow for that visualization/traversal options. Just my two cents on that, Best regards, Armando *Da:*Juan Antonio Pastor Sánchez [mailto:pastor@um.es <mailto:pastor@um.es> ] *Inviato:* mercoledì 16 settembre 2015 02:21 *A:* public-esw-thes@w3.org <mailto:public-esw-thes@w3.org> *Oggetto:* UNESKOS Vocabulary and 2nd SKOS version of UNESCO Thesaurus Dear all, I want to communicate the publication Vocabulary UNESKOS that complements certain aspects of SKOS [1] and ISO-THES [2]. This vocabulary is designed in the context of the proposed UNESKOS, more specifically for the SKOS representation of the UNESCO Thesaurus. The document describing the vocabulary is available at: * http://skos.um.es/TR/uneskos The RDF vocabulary is available for download and use from the UNESKOS namespace: * http://purl.org/umu/uneskos# <http://purl.org/umu/uneskos> <http://purl.org/umu/uneskos> Likewise, the 2nd SKOS version of the UNESCO Thesaurus is available at: * http://skos.um.es/unescothes Includes following features: * Persistent and Dereferenceable URIs. * Turtle and RDF/XML Datasets avalaible for download. * RDFa markup. * SPARQL Endpoint. * Content negotiation avalaible in N3, JSON-LD, etc... This second version makes use of SKOS, ISO-THES and UNESKOS. Along the coming weeks new features navigation within the HTML version will be added. Please, for any suggestion or correction you can contact me. Best regards, Juan -- Juan Antonio Pastor Sánchez, Ph.D. Dep. of Information and Documentation Faculty of Communication and Documentation University of Murcia phone: +34 868 88 7252 <tel:%2B34%20868%2088%207252> http://webs.um.es/pastor pastor@um.es <mailto:pastor@um.es> <mailto:pastor@um.es <mailto:pastor@um.es> > -- Osma Suominen D.Sc. (Tech), Information Systems Specialist National Library of Finland P.O. Box 26 (Kaikukatu 4) 00014 HELSINGIN YLIOPISTO Tel. +358 50 3199529 <tel:%2B358%2050%203199529> osma.suominen@helsinki.fi <mailto:osma.suominen@helsinki.fi> http://www.nationallibrary.fi -- Dr. Juan Antonio Pastor Sánchez Dep. de Información y Documentación Facultad de Comunicación y Documentación Universidad de Murcia Tel: +34 868 88 7252 http://webs.um.es/pastor pastor@um.es <mailto:pastor@um.es> Juan Antonio Pastor Sánchez, Ph.D. Dep. of Information and Documentation Faculty of Communication and Documentation University of Murcia phone: +34 868 88 7252 http://webs.um.es/pastor pastor@um.es <mailto:pastor@um.es>
Received on Wednesday, 16 September 2015 14:35:47 UTC