- From: ashok malhotra <ashok.malhotra@oracle.com>
- Date: Mon, 23 Nov 2009 05:48:13 -0800
- To: Jonathan Rees <jar@creativecommons.org>
- CC: Larry Masinter <masinter@adobe.com>, "www-tag@w3.org" <www-tag@w3.org>
I think that metadata will have as many models as the types of resources that it describes. For example, if you have a Relational Database, there is a special information schema to describe the tables. So, how about saying the RDF is the *syntax* in which different metadata models are described. We will need to specify how different models are described in RDF syntax. In some cases there is ongoing work in this area in others we will need to start appropriate efforts. All the best, Ashok Jonathan Rees wrote: > I'm with you... RDF per se has little to do with models of anything, > any more than XML or ASCII does; it's a way of *expressing* models > syntactically, which is the easy part. (RDF semantics is also helpful > discipline, but also brutally neutral.) You still have to create > vocabularies (ontologies) that do what needs to be done. > > The consumer use cases are the interesting part of the story - linked > data isn't much good if no one's using it - and I think they should be > sought out and/or developed. > > Jonathan > > On Sun, Nov 22, 2009 at 5:23 PM, Larry Masinter <masinter@adobe.com> wrote: > >>> * Metadata model: what is the "data model" for typical metadata applications - >>> the datatypes of the endpoints? >>> The model is RDF. We recommend that all metadate be encoded as RDF. >>> >> RDF 'has' a data model -- things you can say. The question remains, I think, >> whether it is useful, productive, and appropriate to allow "anything you can >> say in RDF" to also be said in metadata. I think the requirements for >> metadata processing may mean that some relations have a much more restricted >> domain. >> >> >>> Metadata in other formats e.g. RDDL, should be translatable into RDF, or >>> encapsulated in a RDF wrapper. >>> >> It's going the other way that is also important. Imagine an audio player >> (WinAmp, iTunes, Windows Media Player) in which you had not just title >> and artist and duration, and so on, but allowed any of those to be >> arbitrary RDF assertions. I think the media player would suffer if it weren't >> possible to restrict the data model of "artist" to be arbitrary rather >> than the dc:creator. >> >> >> >>> * Metadata serialization: how can metadata be encoded in a representation system, >>> be it RDF or something else >>> Metadata is serialized using standard RDF serialization. >>> >> Yes, RDF is one serialization. >> >> >> >>> * Metadata vocabularies: what are appropriate vocabularies for describing various >>> media objects and network services? What is the process by which new vocabularies >>> can or should be developed, described, extended or changed? >>> There exist RDF vocabularies for several domains. Others need to be created. >>> >> I think it's easy to create vocabularies; the real difficulty is vocabulary >> mapping and also the scalability of metadata when merging metadata from multiple >> sources. >> >> >>> * Metadata linking: What are the various ways in which metadata can be associated >>> with "data" or other resources? Link relationships, protocol elements, mechanisms >>> for embedding metadata in various kinds of data? >>> I think this is issue 62: http://www.w3.org/2001/tag/group/track/issues/62 >>> >> Issue 62 focuses on one way of linking; I don't think it is or should be the >> only way. >> >> Larry >>
Received on Monday, 23 November 2009 13:51:07 UTC