- From: Tadej Stajner <tadej.stajner@ijs.si>
- Date: Mon, 29 Apr 2013 10:10:13 +0200
- To: Sebastian Hellmann <hellmann@informatik.uni-leipzig.de>
- CC: Hugh Glaser <hg@ecs.soton.ac.uk>, Denny Vrandečić <denny.vrandecic@wikimedia.de>, semantic-web at W3C <semantic-web@w3c.org>, Felix Sasaki <fsasaki@w3.org>, David Lewis <dave.lewis@cs.tcd.ie>
Hi, all, yes, the examples from Sebastian is what we intended to support with these attributes: ident-ref is when we want to say "this fragment represents that exact resource", and class-ref when we want to say "this fragment represents an instance of that class". To answer some tasks from the previous messages in the thread: * We have a reference demo web service for generating these annotations: http://enrycher.ijs.si/mlw/ * We also have an API to generate them - just use HTTP POST on the endpoint, for example: curl -d "<p>Welcome to London</p>" http://enrycher.ijs.si/mlw/en/entityType.html5its2 * The unfortunate attribute names its-ta-ident-ref and its-ta-class-ref were results of several re-namings and simplifications until we settled on a common consensus that is understandable between use cases of semweb, content management, localization, linguistics, text mining and terminology management. While we don't seem to be winning beauty contests :), in the end we're happy that we have a one-attribute mechanism to solve a meaningful use case for all the parties in the loop. best, -- Tadej On 4/29/2013 7:14 AM, Sebastian Hellmann wrote: > Hello Hugh, > indeed, what you want goes more into the direction of ontology > learning and actually, I would say that your example is arguable. > > If it were a definition, the kind of annotation you propose would be > feasible: > > The terms <span > its-ta-ident-ref="http://nerd.eurecom.fr/ontology#Location">location</span> > and place in geography are used to notice and or identify a point or > an area on the Earth's surface or elsewhere. > > Actually, what you mean and what is practical and useful is: > <span its-ta-ident-ref="http://dbpedia.org/resource/Dublin" > its-ta-class-ref="http://nerd.eurecom.fr/ontology#Location">Dublin</span> > is a location. > > Although the first case, where you can define an ontology and mark up > the references of classes in the text, is nice, it remains rather > academic and marginal, while the other one is relevant for web-scale. > > All the best, > Sebastian > > > Am 28.04.2013 11:48, schrieb Hugh Glaser: >> I'm not sure its-ta-class-ref is what I meant. >> Your example (simplified a little): >> <p><span >> its-ta-class-ref="http://nerd.eurecom.fr/ontology#Location" >> its-ta-ident-ref="http://dbpedia.org/resource/Dublin">Dublin</span> >> is the capital of Ireland.</p> >> Sort of thing I meant: >> <p><span >> its-ta-ident-ref="http://dbpedia.org/resource/Dublin">Dublin</span> is a >> <span >> its-ta-ident-ref="http://nerd.eurecom.fr/ontology#Location">Location</span> >> in Ireland.</p> >> >> I am guessing I would need to do that, rather than use its-ta-class-ref. >> (Assuming you can guess what I think I might be achieving!) >> One says that this is a resource from a class, and the other says >> that this is a class. >> >> >> On 27 Apr 2013, at 19:13, Sebastian Hellmann >> <hellmann@informatik.uni-leipzig.de> wrote: >> >>> Hi Hugh, >>> >>> Am 27.04.2013 18:47, schrieb Hugh Glaser: >>>> Actually, your example <span >>>> its-ta-ident-ref="http://usefulinc.com/ns/doap#developer">someone >>>> who works on</span> is quite interesting as #developer is an >>>> rdf:property. This might actually be problematic later in RDF as it >>>> causes OWL Full, when used as an object. >>>> Ah - I think that is why I put it in - to see what happened :-) >>>> I was thinking of putting a Class in as well, but I guess that >>>> makes less difference. >>> Classes are tackled with its-ta-class-ref . Named Entity Recognition >>> and Linking (i.e class (Person, etc.) and entity link) are a much >>> more common use case than relation extraction, which is why we >>> included it from the start. This was a given separation done by >>> language tools, any how. Making a distinction between instances, >>> properties (object, datatype), classes and annotations is OWL >>> specific, so the motivation+rationale comes from a different domain. >>> -- Sebastian >>> >> >> > >
Received on Monday, 29 April 2013 08:10:49 UTC