- From: Phil Dawes <pdawes@users.sourceforge.net>
- Date: Sat, 17 Apr 2004 15:28:11 +0100
- To: "Seaborne, Andy" <andy.seaborne@hp.com>
- Cc: www-rdf-interest@w3.org
Seaborne, Andy writes: > Phil, > > A nice message to find in all the junk from a week on vacation ... > And yours is a nice one to get back from my holiday from as well :) > > From: Phil Dawes <> > > > > In particular, I couldn't find any way to do optional queries, so > > Veudas has to do multiple queries to fetch optional labels and schema > > information. (the main view panel requires 7 queries to Sesame's 2). > > Have you tried using the "fetch" query langauage? Its appraoch is very > different from RDQL or SeRQL. The idea is to ask a knowledge base for what > it knows about <x>. What is returned is up to the server - the client does > not control the results in the same way the matching subgraph of Joseki or > the rules based CONSTRUCT of SeRQL does. The client gets back a model and > then looks inside to see what's returned. > > The Joseki server can be controlled with differenbt "fetch" handlers. There > is a common one, the bnode closure from a resource, but in general I see > these are being dataset specific. > I had assumed that the fetch model just did bnode closure in a similar way to Patrick's RDFQ 'concise bounded descriptions'. I will have to look at it a bit closer. To be honest, the joseki fetch idea fits in more with the way I see veudas going. I particularly like Patrick Stickler's ideas about how the semantic web should work (at least in bootstrap phase), with conventions for obtaining information about a URI from the URI itself. I'd like veudas to have a number of ways of obtaining information about a resource (e.g. MGET, 301-Redirects, RDQL/SeRQL on well-known stores), and to amalgamate the information for the user on one page. > > Also, I couldn't find any way to do case-insensitive string queries, > > so the label search function is case sensitive for Joseki. > > I may have missed something, so if anybody knows how to do either of > > these things please drop me a mail. > > There is the =~ operator which does regex matching perl style, including //i > for cse insentive matching. > > WHERE (?x foaf:name ?z) > AND ?z =~ /smith/i > Magic - many thanks for this. > > > > Also in this release is the ability to swap between stores of > > different types - e.g. you can switch from a sesame to joseki or > > cgi-file store and back again in the same session. RDF Stores are > > configured via an XML file (which is different from the previous > > release, so it's worth looking at the docs). > > I found using RDF/N3 for config files very good. Not its (N3) design goal!. > I'd like to use N3, but I don't know of a javascript N3/turtle parser. XML is simple hack to solve this problem at the moment. Thanks again for replying, Phil
Received on Sunday, 18 April 2004 06:23:37 UTC