Tuesday, May 30, 2006

Battle Between Two Towers Continues

"Position Paper: A Comparison of Two Modelling Paradigms in the Semantic Web" which continues some of the themes from "Semantic Web Architecture: Stack or Two Towers?".

"One paradigm is based on notions from standard logics, such as propositional logic, first-order logic, and Description Logics...the Classical paradigm...The other paradigm is based on notions from object-oriented databases... and rule languages...the Datalog paradigm."

"The Semantic Web is a very hostile environment for the unique name assumption. There are many and varied sources of information in the Semantic Web, even in the same area, and these sources are free to coin their own identifiers (IRIs) for anything they choose. For example, there are many providers of FOAF information, each of which may choose to use different identifiers to identify to the same individuals."

"FOAF mailboxes form a unique identifier for members of Person (i.e., mbox is an inverse functional property in FOAF). So, if one information source includes:
mbox(Bill Jones, "mailto:Bill Jones@ex.com") and another includes
mbox(William Jones, "mailto:Bill Jones@ex.com"), then it can be inferred from the two sources that Bill Jones and William Jones identify the same individual. This is not possible in the Datalog paradigm..."

"...mary and john could be the same, and due to the cardinality restriction it is inferred that they are, indeed, the same...here is an easy solution—simply state that these two individuals are different...it is also a good idea in general to explicate
inequalities (and disjointness) where they are known."

"Another way in which to apply a local closed world and/or unique name assumption would be to augment the Classical paradigm with constructs that could provide a Datalog-like flavour for portions of the Semantic Web. For example, there is no conceptual problem in providing constructs that state that certain information sources abide by the unique name assumption or are complete in some way."

"A promising direction for the future is to add epistemic constructs to OWL. An epistemic Description Logic provides a formalism that is mostly open, but that can close certain areas of information as desired."

I wonder if applying ideas like Date's SUMMARIZE operator whether things like COUNT, SUM, etc. could be applied to the Classical model.
Post a Comment