> The same XML (or HTML) document with different character encoding can > end up in different byte-streams. Hence, if there are two identical > ontologies backed up at different locations but served with different > character encoding, would you consider they "lied" if they say that they > are the same thing by owl:sameAs? Or if one document has a "meaningless" > empty space added at somewhere in one of the document, you would > consider they lied too about making the assertion?
It all depens on how an information resource is defined. If we choose byte-identity as a criterion for sameness of information resources (and there are good arguments for doing so), then yes, an owl:sameAs statement would be erroneous in such a case. If we want to make the statement that two information sources are not fully identical, but are similar in some meaningful way, we can create an OWL property for that (e.g. in the bio-zen ontology, this property is called "alternative representation"). > So no > matter what you propose, there is always a possibility that some URIs > may break. Hence, your solution will not be a complete solution, will it? Of course there can never be a complete solution, as broken and outdated URIs/URLs can never be fully avoided in a distributed system like the Semantic Web. However, we are trying to reduce the amount of such errors and enable people to correct mistakes when they occur. We are also trying to represent this information in a way that is ontologically correct and less prone to misunderstandings, e.g. the popular confusion of information resources, non-information resources and strings that can be used to retrieve data through some resolution mechanism (which we have witnessed in this discussion once again, by the way). cheers, Matthias Samwald -- "Feel free" - 5 GB Mailbox, 50 FreeSMS/Monat ... Jetzt GMX ProMail testen: http://www.gmx.net/de/go/promail?ac=OM.GX.GX003K11711T4781a
