Re: [legalruleml] Specifications and Namespace

Dear Harold, dear Adrian, and All,

many thanks for your email and for your precious contribution.

We see three interesting issues in your email: goals of LegalRuleML,
methodology of work in the OASIS TC, content.

1. Concerning the goal of the LegalRuleML TC.
The goal in the charter of LegalRuleML is to provide at the end a set of
XML specifications that is able to represent the requirements of the
legal domain as an extension of RuleML. In these months we focalized too
much our attention on RuleML compliance rather than to follow the needs
of the legal community.

The AI&LAW community needs a vertical standard that is easy to use,
conceptually close to the legal domain specificity, human readable,
representative of the legal text. It is so true that LKIF-rule (Tom
Gordon) has been proposed to address such issues (even if it lacks many
important aspects)

Sometime RuleML is too much neutral, it is very verbose and it makes
difficult to express what is clearly expressed in the legal text.

Specific operators for legal domain should make more clear the
representation, help the legal expert to model norms. For me the XML
schema, like in Akoma Ntoso, is, among the other puroposes, a didactical
instrument for leading the norm representation and modelling.
RuleML does not have these charateristics.

So our goal is to develop:
- legal defeasibility operators (e.g., override)
- legal deontic (e.g., obligation, permission, violation, etc.)
- legal temporal model (e.g., interval of efficacy of the norm)
- legal metadata (e.g., source, references, author, etc.)
- legal meta-rules in legal domain (e.g., modification, or atoms that
call other rules)
What we would ike to do is to model also meta-rules that we have not
faced so far, but they were included in the manifesto and also in some
papers by Guido and me.

2. Concerning the method of the OASIS TC.
We have also the duty to not cannibalize the OASIS LegalRueML TC.
Remember that the OASIS TCs IPR specifies: what we have discussed inside
of the TC belongs to the TC namespace. So we think that override, key,
keyref belongs to lmrl in principles, apart the namespace policy that we
will decide to adopt (e.g. to use the same tags in LegalRuleML TC and
also in RuleML, why not, without any namespace for both).

3. Concerning the content of the LegalRuleML.
The refined schema is ok for us (if you remind we discussed it together
in a thread) but we want to refine it again.

Defeasibility could be generic, but also verticalized for legal domain
(e.g., Guido papers).
Temporal issues have to split in two parts jointly coordinated.


                                 ---- LegalRuleML ----
                                /   /      |      \    \
             -------------------   /       |       \    LegalRuleML Metadata
            /                     /        |        \      \      \        \
      Deontic                    /         |         \     sources authors normsQualification
     LegalRuleML  Defeasible RuleML     Temporal      \       \
           \           |  \           LegalRuleML      \     jurisdiction
            \          |   LegalRuleML/         \       \
             \         |  defesibility       Temporal RuleML
             \        |            /
               \       |    LegalRuleML            \       \
                           Argumentation           \       \
                \     |          /                  \       \
                Deliberation RuleML                 Reaction RuleML


So the idea is to develop the temporal model in two directions:
1. one for the norms (e.g. efficacy, enforciability, etc.)
2. and one for the events that are subject of the norms (e.g. if X make
assaults to Y for two months, X is guilty of "stalking crime", if Y
belongs to X and X is a earthquake zone, then Y no tax payments till
15/5/2013).
The two goals are really connected in legal domain and we need to face
them together.
The following example clarifies the concept.
Section 25 of Terrorism Act 2006 has a natural interval of efficacy by
default.
This new provision coming from the Order 2010 suspend this efficacy.
"Disapplication of section 25 of the Terrorism Act 2006
2.  Section 25 of the Terrorism Act 2006 is disapplied for a period of
six months beginning with the coming into force of this Order. "
We have to manage also retroactive events and the event calculus model
is not appropriate.

In this case we need:
a) a method for modelling the natural interval of efficacy (permanent
interval of time)
b) a method for modelling the modification inside of the norm
c) a method for dynamically recalculate the efficacy interval and
readjust the permanent interval of time

Also the "jurisdiction" is an example of what we would like to do.
"Jurisdiction" topic is composed by two very different aspects:
- there is a natural "jurisdiction" of the act itself (e.g. Italian
Constitution is by default under Italian jurisdiction, a UN resolution
is word wide usually). We can for now call "sovregnity" (I don't like
the term, we need to find a better term), just for distinguish it from
the others meaning. This "soveregnity of the act" and it is a property
of the act and it is usually quite a static attribute (EU for the
directives, UN for the resolutions, etc.)
- there is a "court of competent jurisdiction" of the court for the
trial that need for sure a deliberative rule description because it is a
content of the law (e.g. Consumer Code in EU: if X is eCommercCustumer
and ((X is a resident of Y or X lives in Y) and Y is an EU country),
then jurisdiction is EU).

We believe that it is now time to focus on the temporal aspects of
LegalRuleML.
In the next TC meeting we will put these topics in the agenda.

I hope this clarify the goals, the expected outcomes and the methodology
of the LegalRuleML.

Best regards,
Monica and Guido
Il 06/07/2012 00:18, Boley, Harold ha scritto: