- From: Peter Ansell <ansell.peter@gmail.com>
- Date: Tue, 26 Mar 2013 09:29:20 +1000
- To: Oliver Ruebenacker <curoli@gmail.com>
- Cc: David Booth <david@dbooth.org>, Pat Hayes <phayes@ihmc.us>, Alan Ruttenberg <alanruttenberg@gmail.com>, Jeremy J Carroll <jjc@syapse.com>, Umutcan ŞİMŞEK <s.umutcan@gmail.com>, Kingsley Idehen <kidehen@openlinksw.com>, "public-semweb-lifesci@w3.org" <public-semweb-lifesci@w3.org>
On 26 March 2013 06:02, Oliver Ruebenacker <curoli@gmail.com> wrote: > Hello David, > > We agree that there are different interpretations. But you haven't > shown that the boundaries between interpretations are graphs > boundaries (others, including me, think that each interpretation is > global). It doesn't need to be strictly at the boundary of a particular graph. It could also be at the boundary of any set of merged graphs, which in your case would be an hypothetical set containing all existing, past, and present RDF graphs. People are in practice using the closed-world-assumption to work effectively with limited sets of RDF graphs. Whether they have downloaded them all locally into a single graph or somehow restricted their queries to use a set of graphs from various SPARQL endpoints is irrelevant to the results of their queries. This is of course not to mention that even if one keeps the open world assumption and uses OWL, that the set of graphs is limited by explicit owl:imports statements. Hence, effectively, interpretations in OWL (or OWL-DL at least) are, and always were, limited to the imported ontologies from a given base ontology in the view of an OWL Reasoner. Whether they match to one or more RDF graphs is beside the point, as the imports are not necessarily the only place that one could find statements about particular things. However, they are the only places that OWL defines them to exist, so they are effectively limited in this way. (I know that RDF and OWL are defined separately, but they are used together so we need to pragmatically make them work together) Ontology authors regularly redefine upper-ontology terms locally in their ontologies to avoid importing the entire ontology if it would not fit with their desired OWL Profile (or would slow the system down too much). In most cases they will not change the essential meaning of the term, but they may not include all of the statements that were made anywhere about the term as that may not fit their application. If one uses the same OWL principles in their SPARQL queries across raw RDF databases (checking that there is a definition for each term using ?s rdf:type ?type or something similar) they are effectively operating in the same manner. They could source different definitions for a URI across different executions of the query by limiting the query to different graphs. Whether any of these strategies are robust in the long term is still debatable, but there is a distinct possibility that they are more robust than a system that requires a fixed interpretation for all time across all graphs. In practice it is an iterative data management problem, not a fixed true/false logic problem. RDF will not be useful without human data managers to manually verify interpretations and compile realistically useful RDF graphs, even if they have to resort to reinterpreting a definition locally to make things work. Cheers, Peter
Received on Monday, 25 March 2013 23:29:51 UTC