- From: Richard Light <richard@light.demon.co.uk>
- Date: Thu, 30 Jul 2009 08:12:30 +0100
- To: Ryan Shaw <ryanshaw@ischool.berkeley.edu>
- Cc: Yves Raimond <yves.raimond@gmail.com>, Pat Hayes <phayes@ihmc.us>, David Canos <davidcanos@gmail.com>, public-lod <public-lod@w3.org>
In message <af556a820907292317v245191b3xf51630a0e34e4636@mail.gmail.com>, Ryan Shaw <ryanshaw@ischool.berkeley.edu> writes > >We discuss the CIDOC CRM extensively in our tech report, which I will >post a link to here as soon as it is available. I personally am of the >opinion that it is overengineered for Linked Data purposes. But I am >willing to be convinced otherwise. In any case, though the CIDOC spec >discusses historical events, I have been unable to find any examples >of people actually using it to model historical events (a recent post >to the CRM-SIG mailing list asking for examples turned up nothing). I have been thinking about how the CRM might be used to express museum catalogue records as Linked Data, but haven't yet reached a firm conclusion. There is a tension between the "metadata" approach you get in Dbpedia and the CRM approach. In Dbpedia, most of the assertions have the resource in question as their subject, so you end up with a set of independent assertions. Conversely, the natural result of applying CRM modelling is that you end up with a rich and complex structure which isn't at all amenable to SPARQL-style querying. My current view is that it is worth throwing events into the mix, since that lets you express co-contextual assertions ("this object was used by William Wordsworth, at Goslar, in 1799") which would otherwise have to be independent ("this object was used by Wordsworth", "this object was used in Goslar", "this object was used in 1799"). However, for Linked Data purposes you probably get diminishing returns if the assertion structure becomes more complex than this. It certainly ought to be the museum community which eats its own dogfood in this regard, and I'll let the list know of any examples which I get to hear about (or produce!). Richard -- Richard Light
Received on Thursday, 30 July 2009 07:13:17 UTC