ISWC2005 Notes: Uncertainty Reasoning for the Semantic Web 1
I'm going to try blogging the conference like I did for the Web 2.0 one last month. I think it'll be harder here since the topics are very technical, but here goes...
I've just arrived at the conference and joined the Uncertainty Reasoning for the Semantic Web workshop. I'm generally interested in uncertain reasoning and have done some investigation in the context of genealogical information. I'm also expecting this workshop to be useful in light of Justin's thoughts on quantum superposition of concepts.
Mauro Mazzieri is speaking about A Fuzzy Semantics for RDF. He says there's a need for fuzzy data because of the vagueness of natural language and also trust metadata. His approach is to associate a value outside of the domain to each triple, extended n-triples:
n: s p o . where 'n' is the trust value
s p o . is equivilent to 1: s p o
Oops this is followed by lots of RDF model theory that goes over my head. I think the presenter is saying that in traditional RDF, an interpretation is only true if all the triples are satisfied. In a fuzzy RDF, it's possible to find interpretations that are true but only satisfy some of the triples - those with a trust value above a specified amount.
A class is a fuzzy set of resources. A similar extension can be made to OWL, from fuzzy logic to fuzzy description logic (an upcoming topic I think)
Phew! The next speaker is Paulo Costa on PR-OWL: A Bayesian Ontology Language for the Semantic Web (paper)
The problem: the semweb is intended to provide a means to express and reason about diverse forms of knowledge. Uncertainty is ubiquitous in our world thus representing uncertainty is essential to the semweb.
What is needed: a probabilistic knowledge base and probabilistic ontologies. Usual definition is that an ontology is an explicit, formal representation of knowledge about a domain of application. However, a probabilistic ontology also includes statistical regularities that characterize the domain; inconclusive, ambiguous, incomplete, unreliable, and dissonant knowledge related to entities of the domain; and uncertainty about all the above forms of knowledge.