Hi Paul, hi David, first of all, @Paul: Could you please check what's wrong with my changes to the build.xml? I don't see http://www.openmath.org/cd/arith1.rdf. However, when I run ant locally (disabling those XSLT tasks that don't work locally because of the XSLT 1.0 vs. 2.0 issue I mentioned previously), the *.rdf files are created in target/site/cdfiles2/cd. I don't see that any other change to build.xml is required, besides creating the desired files in some subdirectory of "target/site".
Therefore I assume that something goes wrong on the server. Could you possibly show me the output or logfile that you are getting when you try to regenerate the homepage? In replying to your (David's) mail, let me start with the most important issue: 07/23/2011 12:40 AM David Carlisle: >> Now there is a little remaining challenge: The stuff should be >> published at the canonical URIs like >> http://www.openmath.org/cd/arith1#plus, so that it becomes >> retrievable, but so far I have published it at a different address >> just for getting the first step done. > > those URIs used to work (and examples of them were in the MathML2 spec) > so needed to link check to get to recommendation status. Going to the > html version. They broke at some point when we (I) dropped the ball on > the old cocoon based website. It would be good to get them back if we > could I have no idea what's wrong. @Paul, do you? Concrete URLs like arith1.ocd and arith1.xhtml work (just arith1.rdf does not yet) – but arith1, which used to work, indeed no longer works. However, the dot_htaccess file in the svn defines a redirect. So something must be wrong either in the central Apache configuration, or when copying dot_htaccess into some actual .htaccess on the server. > (it would still be possible to serve the rdf from the same url > given a bit of content negotiation apache voodoo on the server) Exactly, and locally I have already figured out how that would work. With Paul's help I will apply it on openmath.org as soon as the *.rdf files get copied there. In a nutshell it works as follows: * If the client requests text/html or application/xhtml+xml from a URL without filename extension, we make a 303 redirect (that's the common convention) to the respective *.xhtml file * If the client requests application/rdf+xml, we redirect to *.rdf * If the client requests application/openmath-cd+xml, we redirect to *.ocd If the client doesn't request anything (e.g. when using wget as a client) we are free to serve anything. It's basically up to discussion what the default in that case should be – maybe *.ocd? > Interesting/ Do you know if the RDF property URIs used here are > compatible with those used here > > http://monet.nag.co.uk/monet/publicdocs/ontologies/openmath.owl The URIs are "compatible" in that both MONET and my data use the canonical OpenMath URIs, plus, that I make RDF available right from these URIs. > in the monet project (If I recall correctly, it's been a while:-) > we didn't attempt to convert individual properties to rdf, and stopped > at just defining the symbols, but it might be good to use the same uris > for the same things or at least owl:sameas declare they are the same, or > just explicitly retire the monet ones, or something. Therefore, any such explicit action is not necessary. In terms of how the CD structure is modelled, my approach is intentionally incompatible with MONET's. MONET basically says that each symbol is an OWL class. I was rather interested in modelling, as OWL classes, the abstract model of an OpenMath CD, and thus in being able to state things like "a CD defines symbols, which may have mathematical properties or examples". In my model there is a class for symbols, and, e.g., arith1#plus is an _instance_ of that class. IIRC MONET's choice of modelling symbols as classes was caused by technical limitations of the storage/reasoning software used at that time (http://monet.nag.co.uk/monet/publicdocs/monet_onts.pdf). Theoretically, it's problematic my data declare each OpenMath symbol to be an instance, whereas the MONET data declare them (using the same URIs) to be classes. That may create problems with OWL reasoners. On the other hand, the possibility is very low that any tool/crawler/engine/whatever will ever see the MONET data, as they are not retrievable by requesting application/rdf+xml from openmath.org (no blame on anyone – that technique was not yet known in 2004), but need to be downloaded manually from monet.nag.co.uk. If the MONET data contained any information that my data didn't, that _information_ would definitely have been worth preserving (e.g. by integrating it into my data) – but they don't contain any such information. They contain a subset of the full information of the CDs (which I'm trying to preserve as faithfully as possible), just modelled in a different way. Cheers, Christoph -- Christoph Lange, Jacobs University Bremen http://kwarc.info/clange, Skype duke4701 Workshop: Ontologies come of Age in the Semantic Web (OCAS) Ontology User Experience Challenge (1st prize US$ 2000) October 23 or 24 at the Intl. Semantic Web Conference, Bonn, Germany Submission deadline August 15, http://ocas.mywikipaper.org _______________________________________________ Om mailing list [email protected] http://openmath.org/mailman/listinfo/om
