On Wed, Apr 24, 2013 at 12:28 PM, Alessandro Adamou <[email protected]> wrote: > On 23/04/2013 14:51, Rupert Westenthaler wrote: >> >> On Tue, Apr 23, 2013 at 1:52 PM, Alessandro Adamou <[email protected]> >> wrote: >>> >>> It would be enough for me to have owl:sameAs links across entities in the >>> DBpedia namespace, non-interlanguage and not necessarily transitive. >> >> redirect links - even interlanguage one - are present but do use the >> rdfs:seeAlso property. > > > Well there are redirect datasets such as > http://downloads.dbpedia.org/3.8/en/redirects_en.nt.bz2 (this one uses > dbpedia-owl:wikiPageRedirects)
dbpedia-owl:wikiPageRedirects are present in the index. However those links are mapped to rdf:seeAlso. > But come to think of it, perhaps DBPedia is not managing these subtleties > yet? redirects are present, because the information a users is searching is present on the other page. Because of that there are a lot of redirects to sub-sections of other pages. Most likely this is also the case in the example you gave. > >> I was thinking you refer to owl:sameAs links to geonames, freebase, ... > > > That's coming next, but first I need to resolve ambiguities in DBPedia. I'm > writing an autocomplete functionality similar to the VIE one, but it should > present a single entity only once (and in a very precise way). If I get 10 > DBPedia entities and nothing telling me they are (strongly) the same, > there's not much of a point in going to other datasets. The EntityLinking engine does the same. With the "Redirect Field" and "Redirect Mode" you can tell the engine to not use the extracted Entity, but follow a reference to the real one. This allows the EntityLinking engine to follow redirects in the case of DBPedia. It would be also possible to build an index that has only a single Entity and add all redirected Entities as alternate labels. I even created such an index once, but discovered that is performed very bad for entity linking. Mainly because Entities with a lot of alternate labels got down ranked because of the Solr length norm. But with STANBOL-1027 resolved one could give this an other try. best Rupert > > Thanks, > Alessandro > > > >>> I'm not sure if there is a specific dump of what I'm looking for at >>> >>> http://downloads.dbpedia.org/3.8/en/ >>> >>> iri_same_as_uri_en is only part of what I'm seeking. >>> >> I have never understood for what this is useful. >> >> best >> Rupert >> >> >>> But anyway, if I were to find it and wanted to load it on my own Stanbol, >>> the reference documentation to follow is still the one on >>> 'entityhub/indexing/dbpedia/README.md' right? >>> >>> Cheers >>> >>> Alessandro >>> >>> >>> >>>> Hi Alessandro, >>>> >>>> The dbpedia index used on dev.iks-project.eu does not contain >>>> owl:sameAs information. For those to be present one would need to >>>> import the according RDF dump files (by copying them to the >>>> 'indexing/resources/rdfdata' directory. >>>> >>>> If you make an proposal what links would be interesting we could add >>>> those files to the default configuration of the dbpedia indexing tool >>>> >>>> best >>>> Rupert >>>> >>>> On Mon, Apr 22, 2013 at 9:54 PM, Alessandro Adamou <[email protected]> >>>> wrote: >>>>> >>>>> Hi, >>>>> >>>>> I was fiddling around with the demo installation at >>>>> http://demo.iks-project.eu/stanbolfull querying the entityhub for an >>>>> application I'm working on. Before diving deep into investigation I >>>>> just >>>>> wanted to ask: is that installation indexing owl:sameAs coreferences at >>>>> all? >>>>> >>>>> I've been POSTing on >>>>> http://demo.iks-project.eu/stanbolfull/entityhub/sites/find with the >>>>> following: >>>>> >>>>> * name = "gaddafi" >>>>> * limit = "20" >>>>> * ldpath = "same = owl:sameAs :: xsd:anyURI ;" (urlencoded) >>>>> >>>>> hoping I didn't botch the trivial LDPath program. Not having much luck >>>>> here, >>>>> I get several equivalent individuals but no values for "same" e.g. >>>>> >>>>> http://dbpedia.org/resource/Colonel_Gaddafi >>>>> http://dbpedia.org/resource/Gaddafi >>>>> http://dbpedia.org/resource/Muammar_al-Gaddafi >>>>> http://dbpedia.org/resource/Muammar_Gaddafi >>>>> >>>>> I am speculating the DBPedia index over there does not include >>>>> owl:sameAs >>>>> links? Have we an index that does? >>>>> >>>>> Of course I could ask sameas.org but I'm more keen on a Stanbol-only >>>>> solution >>>>> >>>>> Best, >>>>> Alessandro >>>>> >>>>> -- >>>>> Alessandro Adamou, Ph.D. >>>>> >>>>> Knowledge Media Institute >>>>> The Open University >>>>> Walton Hall, Milton Keynes MK7 6AA >>>>> United Kingdom >>>>> >>>>> Alma Mater Studiorum - Università di Bologna >>>>> Department of Computer Science >>>>> Mura Anteo Zamboni 7, 40127 Bologna >>>>> Italy >>>>> >>>>> >>>>> "I will give you everything, just don't demand anything." >>>>> (Ettore Petrolini, 1917) >>>>> >>>>> Not sent from my iSnobTechDevice >>>>> >>>> >>> >>> -- >>> Alessandro Adamou, Ph.D. >>> >>> Knowledge Media Institute >>> The Open University >>> Walton Hall, Milton Keynes MK7 6AA >>> United Kingdom >>> >>> Alma Mater Studiorum - Università di Bologna >>> Department of Computer Science >>> Mura Anteo Zamboni 7, 40127 Bologna >>> Italy >>> >>> >>> "I will give you everything, just don't demand anything." >>> (Ettore Petrolini, 1917) >>> >>> Not sent from my iSnobTechDevice >>> >> >> >> -- >> | Rupert Westenthaler [email protected] >> | Bodenlehenstraße 11 ++43-699-11108907 >> | A-5500 Bischofshofen >> > > > -- > Alessandro Adamou, Ph.D. > > Knowledge Media Institute > The Open University > Walton Hall, Milton Keynes MK7 6AA > United Kingdom > > Alma Mater Studiorum - Università di Bologna > Department of Computer Science > Mura Anteo Zamboni 7, 40127 Bologna > Italy > > > "I will give you everything, just don't demand anything." > (Ettore Petrolini, 1917) > > Not sent from my iSnobTechDevice > -- | Rupert Westenthaler [email protected] | Bodenlehenstraße 11 ++43-699-11108907 | A-5500 Bischofshofen
