RE: [BioRDF] URI Resolution

[Matthias Samwald]

> - If the server replies 303 See Other, follow the link in the > response to get information about resource. [obscure hack but worth > a try] > (see http://www.w3.org/2001/tag/issues.html#httpRange-14)   I guess we should find a diplomatic formulation for that.     > - Disadvantage: you need an OWL engine to interpret resolution > information represented in this way, and not all applications have > an OWL engine. [so why not get one and link it in?]   I would prefer a different, much simpler approach to ontology-based URI resolution. I would suggest that the relation between an information resource and the URL that can be used to retrieve a realisation of this information resource is explicitly encoded as a triple, without the necessity of inferencing on the client-side. Furthermore, there should be a 1:1 mapping between the URL and the data that can be retrieved via a HTTP GET (no content negotiation). The URI of the information resource should preferably NOT be identical with the URL that can be used to get a realisation of it. This way, all of the different entities we are dealing with (non-information resource, information resource, URL to get a realisation of the information resource) would be cleanly separated. If a webserver fails or files are moved to another server, the triples can be updated accordingly.   URI resolution for information resources is very important for any application on the Semantic Web, and we need to make it as barrier-free as possible. Requiring OWL inference for such a basic task is, in my opinion, simply too demanding to find widespread adoption. Furthermore, leaving the mapping between URI and URL to the server side frees us from the necessity to standardize the resolution ontology proposed by Alan, which would surely be an arduous task. In the simplest case, we would only need to standardize a single property -- to associate an information resource with a URL.   Xiaoshu, you probably criticise the need for additional triples, but you need to be aware that these additional statements are only made for *information resources*, not for all resources in the RDF graph. If you look at most of the current biomedical RDF/OWL datasets, information resources are only a small fraction of all defined resources.   -- Matthias Samwald