- From: Ian Davis <iand@internetalchemy.org>
- Date: Wed, 27 Jul 2005 16:33:49 +0100
- To: public-web-http-desc@w3.org
I though I'd throw my contribution for web description to the lions: http://iandavis.com/2005/web-description/waif-20050727 This is a description format for web applications that work in terms of URIs, resources and representations. It's a sane profile of RDF/XML, in fact the XML serialization only uses RDF for cross-referencing via URIs. The format itself is constrained by an XML Schema so is processable via normal namespace-aware XML tools. My goals in designing this format have been: * Make it possible to describe applications based on the Web architecture of resources identified by URIs. * Make it possible to describe applications and services and how to interact with them in a simple and intutive fashion. * Make it possible to separate the application interface from the multiplicity of possible implementations. * Make it possible to determine when two syntactically different parameters from different services have the same meaning. * Make it possible to reuse components within and between descriptions. I think I've met those goals with this format and have something that is flexible enough to describe web services such as Amazon's ecommerce REST interfaces and simple collections of resources accessed via GET, POST, PUT etc (otherwise known as web sites :) One of the goals above talks about determining whether different parameters have the same meanings. An example of where this is useful comes from the domain I've recently started working in: libraries. Typically libraries provide access to their catalogs via an OPAC which is a web-based search system. There are no standardised interfaces for OPACs so each vendor implements entirely different mechanisms to perform the same conceptual function. For example searching by ISBN in a Talis OPAC uses a URI like this: http://prismdemoa.talis.com:6080/TalisPrism/doOpenURLSearch.do?isbn=0120078376 whereas, in an Innopac OPAC, the same search uses a URI like: http://library.dur.ac.uk/search/i?SEARCH=0120078376 These are conceptually the same search and in WAIF would be related by using the same URI for the purpose of each parameter (I show this in one of the examples). If I have an ISBN I can use WAIF descriptions to build a search URI for any described service that has a parameter with that purpose. I'm looking forward to any comments, criticism, outright laughter etc. Ian -- http://internetalchemy.org | http://purl.org/NET/iand Working on... Silkworm <http://silkworm.talis.com/>
Received on Wednesday, 27 July 2005 15:34:02 UTC