Jeni's Musings

Hey Jeni. I’m quite a fan of your blog and have been tracking rdfQuery closely lately and hope to use it with FuXi (at the server) to investigate a parallel paradigm for my current approach of ‘XForms in the client, XML on the wire, and RDF in the store (via transformation)’, i.e., one where RDF is bound to markup via javascript and dispatched to inference services (which is where FuXi comes in) and RDF datasets. More on that later (hopefully I can get around to blogging about it).

Anyways, I wanted to respond to your philosophical note:

On a more philosophical note, it strikes me that the concept of aggregation contradicts the Open World assumption. I can only know that the number of members’ winding-up order notices was exactly 30 if I know that I know of all the members’ winding-up order notices that exist. Pragmatically, in many cases this is going to be just fine, because we know that the datasets that we’re using are complete (our World is Closed), but it does slightly concern me that it’s impossible to do much useful data analysis without contradicting one of the fundamental tenets of the Semantic Web.

I think the OWA and its association with the Semantic Web is over stated and actually is myopic. I personally don’t think the Semantic Web will ever be able to break through a certain threshold of adoption and pragmatic usage without support of some form of CWA-based querying and inference. Note that it basically already does in the form of using OPTIONAL/FILTER/!BOUND for the equivalent of negation of failure (which follows the CWA that if it is not in the dataset/base then it is considered false)

Common sense semantics are more inline with the CWA. The problem with the OWA (and the reason why I think weaving it into the philosophy of the SW has done more harm than good) is that it assumes universal truths (despite the fact that the motivation for using the OWA is the avoidance of universal truths). Consider the following two english statements:

Consider the class of people who are guilty that is considered the opposite of the class of people that are considered innocent (where ‘opposite’ has a logical sense and a ‘common’ sense).

In a scenario where negation is used with the OWA. The intuition for this kind of negation (classical negation) is often described as the law of the excluded middle. In this case the class of people who are guilty is defined precisely as everyone minus those who are members of the class of people who are innocent. It is an absolute definition, however, as we know innocence, guilt, etc. are imprecise classifications in real life.

Even with an OWL axioms that states that the class of people who are innocent is disjoint with the class of people who are guilty, we cannot infer a person is guilty without this ‘negative information’. And even with this negative information we need to rule out their innocence in a purely logical/mathematical way.

But, as we know, although judicial systems consider the application of logic as the ideal way to prosecute/defend, inevitably they can only rely on known evidence to help a jury come to a particular conclusion. People are convicted of crimes even without irrefutable evidence.

This is the difference between negation via the CWA and classical negation. In non-nonotonic reasoning (which is often considered common-sense interpretation of negation), if there is no statement P in your database, then you can conclude not P. The emphasis shifts from the application of are purely mathematical process to derive complementary information to the administrator of the database to attempt to collect complete information about P, or at least annotate the database to the effect that he/she cannot do so.

There are many practical situation where we do have complete information about a particular predicate (consider banking, for instance). A knowledge representation that doesn’t support both forms of reasoning about negation will forever appeal only to logicians and not ‘engineers’.

If you are interested in this counter argument, I’d suggest looking at “Negation and Negative Information in the W3C Resource Description Framework” (at least the first part that lays out arguments about the futility of a system that adheres to only the OWA):

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.60.6232 , we cannot infer a person is guilty unless