- From: Giovanni Tummarello <giovanni@wup.it>
- Date: Fri, 04 Feb 2005 10:43:58 +0100
- To: www-rdf-interest@w3.org
Greetings, this is a contribution to a thread from some time ago about "non SemWeb" applications of RDF. By non SemWeb it was meant, i believe, applications to fields that naturally live without a network connection and , if that is the case, could probably make a closed world assumption (or have been doing it so far :-) ). An example of such application is textual encoding, that is, writing in machine processable how a literary text is structured ( pages, rows, paragraphs, sentences, verses.. all the way down to grammatical elements such as "verb" "subject" and so on.). For those interested I'd like to point at embryonic stage paper (warnings: highly questionable grammar, incomplete and likely partially incorrect) which explores textual encoding using SW tools. Motivation: For decades people have been creating their specific data structure, for the problem they needed to solve. When XML came along, applications began making extensive use of it as a data structure format, but it is clear that just a small subset of problems can be naturally represented as a tree. So what happens is that most complex XML based data formats in practice make a large use, e.g, of attributes to be able to construct what they needed from the start: graphs. Trying to apply the semantic web technologies to these sort of problems is very interesting since the payoff is big. If it works, if we can in fact address domain specific use cases and queries with the tools and semantics from the SW then we can immediately: * enjoy the use of a commonly understood set of high level semantic constructs rather than relaying on application specific idiosyncrasies in the data model. This in short means that developers would use constructs they know therefore the project is cheaper to build and less error prone * be able to use a large and growing set of tools and applications * be naturally to expand these applications to a distributed environment The paper: http://giovanni.ea.unian.it/temp/rdftef.pdf from a semantic web point of view, this simple use case i think highlights the shortcomings of the SW query languages that have a fixed path length. The data model that we propose for the encoding makes SparQL (for example) rather useless to compute the example query in the paper (if not with a preprocessing or the addition of a list aware operator as we suggest) . Prolog (or metalog would do) on the other hand does it really sweet :-) P.S. I used "non sem web" since these applications where previously referred to as such, but i also agree that they are , in fact, SemWeb applications in its wider, nicer scope Giovanni Tummarello, also on behalf of Christian Morbidoni
Received on Friday, 4 February 2005 09:45:51 UTC