Re: What is a semantic web application

>>  If a system uses RDF or an agent system or an ontology does that make it
> Far from being an expert, my view is that whenever you start using semantic 
> data (RDF, OWL, etc) you already have a "semantic web" application

even within 'semantic web' there are distinctions between 'upper-case' and 'lower-case'

i can't find a good wikipage arguing what the difference is, but i believe its generally been used to mean whether you used the W3C-annointed layer-cake components or not

lower-case would be something like Atom or tags (textual symbols without corresponding URIs)

then again, i'd consider something like MetaWeb's Freebase an upper-case semantic web platform. since its a directed graph with named links (using URIs), although it doesn't use the W3C-annointed serialization formats and base schemae. clearly one organization's recommendations cant be the difference between lower-case and upper-case, it is likely more some quantum threshold surpassed when you embue enough semantics in an easily accessible way (and reusing JSON is definitely more accessible than necessitating an extra level of parsers/serializers for things like N3 and SPARQL where they werent really necessary. (we don't express CSS in XML or JSON either, clearly someone at W3C has a parser fetish))

using the 'rel' attribute on links and labeling the arc with an ab solute or qname (dc:title) uri is also a good direction to move towards, and is also the foundation formalized in RDFa/eRDF, and is also backwards compatible to webpages made in the dark ages of the web..


the other paradox i see a lot is the fight for complete unambiguity (eg, the use of 302/303 redirects, primaryTopic and similar metaURIs, hair-splitting semantic distinctions between a document/URI/whatever and what the document is representing, etc. eg a good proportion of this list's traffic). and the other side of this is the unlabeled 'tags' which are so popular everywhere, despite their lack of (global) URIs. so theres a definite need for sort of fuzzy-inferencing, and GRDDL for stuff which isnt pure XML. perhaps when you get into this territory and bridge the two worlds you become the all-caps SEMANTIC WEB?

Received on Thursday, 6 December 2007 22:36:42 UTC