|
Hi Tara,
Seems like a productive round of comments, so just a few more responses from me.
Thanks for the note and for the presentation at RuleML. And for the movie recommendation….
Some off-the-cuff (and defeasible) responses to your notes….
I thought I sent this to the LegalRuleML list after the tutorial, but apparently I sent it to the wrong email address:
I got a lot of questions from the tutorial presentation.
There was interest in how LegalRuleML will be adapted to the individual legal systems, and then how would they interoperate.
As I understand it, LegalRuleML has no conception of 'individual legal systems' per se. Rather, all of the contextualisation components should be able to handle that. In addition, I understood the intention to be that laws no matter where would be expressed
in the 'language' of LegalRuleML. Interoperability is intended to be 'guaranteed' by having any law written in the same ML. However, it remains to be seen just how LegalRuleML is used and whether issues arise that require adaptation.
I agree with Adam. Indeed, we have to see if RuleML will be used to determine what adaptations are needed.
I don't think we need to change LegalRuleML at this point, but *if* LegalRuleML is going to be adopted, I think we have to have a reasonable answer to this question.
As Guido mentions below, one aspect of adapting LegalRuleML to different legal systems is different legal ontologies. This suggests that the interoperation between such legal systems may depend on some alignment of these different legal ontologies.
There is alignment between different legal ontologies, and there is alignment of any ontology with LegalRuleML. If two or more ontologies align with LegalRuleML, then there seems to me to be an implication that the ontologies align with one another at
the same places. But, more than that we cannot say, for how the ontologies otherwise treat the classes and individuals is another matter as the properties in the ontology need not be introduced into LegalRuleML itself.
There was a little confusion about the expressivity of an RDF-based syntax - it was necessary to explain that, like OWL in RDF/XML syntax, the expressivity is not limited by the utilization of an RDF metamodel. There was additionally interest in the application
of defeasibility at lower expressivitiies. After some discussion, we concluded that defeasibility is not relevant at the level of RDFS or less unless there is an explicit mechanism for asserting conflicts, which would increase the expressivity of the language.
Don't I understand the issue here.
It's not an issue for LegalRuleML, but the point is that it is not possible to create an inconsistency in RDF, so there are no conflicts to resolve with defeasibility. There also aren't any rules, so it is something of a moot point. I think it is mostly an
example of someone who is primarily interested in ontologies trying to figure out if they need to understand defeasibility.
Moot point. For lots of work on ontologies, defeasibility does not matter; ontological reasoners do not handle defeasibility. However, defensibility can be interesting to someone working with ontologies if (1) they are interested in aligning or combining
different ontologies such that the alignment or combination gives rise to conflicts and priorities or (2) the ontology is dynamic, in which case, it might be non-monotonic.
There was a law student in the tutorial (Christine from Brazil) who is developing a legal ontology . She wanted to know if LegalRuleML would have its own legal ontology. I asked her to send me her papers.
LegalRuleML is not intended to have its own legal ontology, and we were deliberately neutral on this point. Yet, one wonders how long that position can be sustained.
The issue of ontologies is related to different legal systems. For example, the concept of “crime” in the Italian legal system is different form the concept of “crime” in the Australian legal system, and these are defined in Sections of Acts/Codes. Furthermore,
it Italy, the concept of crime is different in criminal law and in civil law, and both concepts are defined by articles of the Criminal Code and the Civil Code.
Perhaps we do need a story here. In a formal ontology, the concept is the set of individuals denoted along with the object and individual level properties (and whatever constraints there are, etc). Informally, the concept is defined by other concepts
in some (legal) code. The example mentioned here about "crime" is the latter, whereas someone thinking about formal ontologies has the former. If a concept like 'crime' means something different in two or more a formal ontologies, then the differences are
grounded in the individuals and properties (etc) of each ontology. But, for our purposes, we will have context around the concept, so 'crime' in LegalRuleML can vary with reference to the legal code. We don't treat concepts in the 'formal ontological' sense.
If we did, then variation would mean that where we have context, we refer to the rest of the definition of the concept in the ontology being used. Sort of makes sense to me at the moment….
Cheers,
Adam
The University of Aberdeen is a charity registered in Scotland, No SC013683.
Tha Oilthigh Obar Dheathain na charthannas clàraichte ann an Alba, Àir. SC013683.
|