Hello, I am Adrian and working with Pascal at the hbz on Linked Open Data stuff, especially lobid.org. I subscribed to this list some days ago.
>>> On 6.5.2013 at 18:15, Sebastian Hellmann >>> <[email protected]> wrote: > Bookmashup came from an earlier time and maybe it makes sense to rethink > owl:sameAs for bookmashup. > They are of class foaf:Document. This sounds reasonable. > The only problem is, that we need a more general property to query such > links. > Use case is to query all outgoing links for all books: > > > Current: > select * Where { > ?s rdf:type <http:dbpedia.org/ontology/Book> . > ?s owl:sameAs ?o > } > > If we start replacing owl:sameAs this gets more difficult: > > select * where { > ?s rdf:type <http://dbpedia.org/ontology/Book> . > ?s ?link ?o . > FILTER ( ! regex(str(?o), "^http://dbpedia.org";)) > } I don't understand why you'd need such a query if ?o is something like <http://lobid.org/resource/HT014469321>. Or do I get something wrong here? > > Pascal, do you know a property, which we can use in addition? > rdf:seeAlso or skos:related or umbel:isLike to these match? rdfs:seeAlso would definitely be an improvement. skos:related has skos:Concept as rdfs:domain and thus is problematic. And umbel:isLike suggests a closer match than there might be. > We could > 1. weaken existing owl:sameAs links > 2. add a domain specific property. I think both makes sense. Weaken owl:sameAs to rdfs:seeAlso and allow adding domain-specific properties. > By the way, a foaf:Document is a frbr:Work or a frbr:Manifestation or > both... ...or might be a frbr:Item or even a frbr:Expression. It shouldn't be two or more of these, though, as according to the general FRBR model they are disjoint. E.g. the FRBR core ontology (one of the different FRBR ontologies) reads: <http://purl.org/vocab/frbr/core#Work> owl:disjointWith <http://purl.org/vocab/frbr/core#Expression>, <http://purl.org/vocab/frbr/core#Item>, <http://purl.org/vocab/frbr/core#Manifestation> . But we better not start a FRBR discussion here. Usually, they absorb much time that is better spent elsewhere and don't lead to anything productive. Having said this, it might be we made a bad choice using these FRBRish properties in lobid.org. But we think, at least the distinction between work, manifestation and item makes sense. Anyway, you should only use of the domain-specific property http://rdvocab.info/RDARelationshipsWEMI/manifestationOfWork to link from DBpedia to library catalogs etc. if you think it makes sense to type literary works on DBpedia as FRBR:Work. I think one can reasonably argue it does make sense. - Adrian > Am 06.05.2013 17:43, schrieb Pascal Christoph: >> Hi Sebastian, >> >> Am 06.05.2013 12:54 schrieb Sebastian Hellmann : >> >>> I merged your links into the repo. Very good, we will include them in >>> the data base soon. >> :) >> >>> I changed the readme to adjust for your requirements. >> I've adjusted[1] a misunderstanding resulting from my incorrectly first post > , >> where I wrongly stated that: >> >> <dbpedia:...> <http://rdvocab.info/RDARelationshipsWEMI/workManifested> > <lobid:...> >> >> To be correct, it should be >> "http://rdvocab.info/RDARelationshipsWEMI/manifestationOfWork";, thus > implying >> that the dbpedia entries are frbr:Works (not frbr:Manifestations). (Our >> provided links, however, were correctly predicated). >> >> [...] >> >>> Additionally, you can include types, which result from inference of >>> the usage of the domain-specific linking property, e.g. the rdfs:domain >>> of the property. E.g. rdrel:workManifested is rdfs:domain >>> rdafrbr:Manifestation, which entails that DBpedia entries should be of >>> rdf:type rdafrbr:Manifestation. >> my fault, see above, it should be: >> "DBpedia entries should be of rdf:type rdafrbr:Work" >> >>> 3. Note that we also count links to other classes as links, so if you >>> want to add an external classification using rdf:type as linking >>> property, that is fine as well. >> hm, we at hbz are hesitant with that. While we don't want to discuss this > here >> and now we want to make you aware of arising problems: >> While we think an encyclopedia generally describes a kind of abstract "work" >> rather than a concrete "manifestation" (e. g. a concrete edition of a book), > we >> find that others classify dbpedia resources in a different way. E .g, if I >> understand correctly, the "bookmashup" dataset seems , like lobid, to > describe >> *manifestations* of books, not *works*. But the bookmashup links with > predicate >> "owl:sameAs" to dbpedia resources. These different approaches might result > in a >> contradictionary assumptions. This is another example why the "owl:sameAs" >> predicate should be used with great care. >> >>> Thanks again Pascal, >>> Sebastian >> thank you Sebastian - looking forward to see links from dbpedia to our lod >> service :) >> >> pascal >> >> [1]https://github.com/dbpedia/dbpedia-links/pull/11/commits >> >>> >>> Am 03.05.2013 16:15, schrieb Pascal Christoph: >>>> Hi *, >>>> >>>> we[1] want to provide links between lobid and dbpedia. We would like to >>>> use > an >>>> other predicate as the recommended ones[2]. >>>> >>>> lobid describes manifestations of library resources, e. g. books. We find > that >>>> predicates like "owl:sameAs" do not fit , because the dbpedia resources > often >>>> don't only describe books but also other forms of manifestations (e .g. a > play >>>> or a movie). Thus we use the predicate rdrel:workManifested[3] to link our >>>> manifestations to dbpedia resources, e. g.: >>>> >>>> <http://lobid.org/resource/HT014469321> >>>> <http://rdvocab.info/RDARelationshipsWEMI/workManifested> >>>> <http://dbpedia.org/resource/The_Last_Man> . >>>> >>>> This "means" : >>>> the lobid-resource is a "physical embodiment of an expression of a work", >>>> and >>>> that work is the dbpedia-resource (while work is defined as "A distinct >>>> intellectual or artistic creation"). >>>> >>>> As for the dbpedia_links we would like to use the inverse predicate - that >>>> would be rdrel:manifestationOfWork[4] . This would imply that these >>>> dbpedia-resources are rdf:type rdrel:Work[5] (we find that quite fitting, >>>> although not every wikipedia entry is actually a "work" (e .g. [6] is >>>> rather > a >>>> "manifestation" - sure someday someone will correct this entry ). >>>> >>>> What do you think - should rdrel:manifestationOfWork be recommended in our > case? >>>> >>>> -o >>>> >>>> [1]http://lobid.org/ >>>> [2]https://github.com/dbpedia/dbpedia-links >>>> [3]http://rdvocab.info/RDARelationshipsWEMI/workManifested >>>> [4]http://rdvocab.info/RDARelationshipsWEMI/manifestationOfWork >>>> [5]http://rdvocab.info/uri/schema/FRBRentitiesRDA/Work >>>> [6]http://dbpedia.org/resource/Five_Go_Off_to_Camp >>>> >>>> ------------------------------------------------------------------------------ >>>> Get 100% visibility into Java/.NET code with AppDynamics Lite >>>> It's a free troubleshooting tool designed for production >>>> Get down to code-level detail for bottlenecks, with <2% overhead. >>>> Download for free and get started troubleshooting in minutes. >>>> http://p.sf.net/sfu/appdyn_d2d_ap2 >>>> _______________________________________________ >>>> Dbpedia-discussion mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion >>>> >>> ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may _______________________________________________ Dbpedia-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
