metamodel/glossary ramblings

[Tara Athan <taraathan@gmail.com>]

Hi, all. I've been trying to incite some discussion on the skype chat, 
but got no takers. I'll embed and extend my skype comments here, for the 
record and for the agenda of the next meeting.

I've just posted the Glossary document in the repo, and I have been 
going through it and the RDFS schemas while preparing my slides on the 
metamodel. The definitions in the Glossary don't all make sense to me, 
especially when considered as the basis of a taxonomy, as embodied in 
the metamodel. Here are some particular questions.

[7/5/13 6:41:15 PM] Tara Athan: In the Glossary, we define Authority as 
"any body with the power to create, endorse, or enforce legal norms" but 
we don't say what a "body" is. Would it be correct to replace "body" 
with the generic "entity", or is there some implicit restriction 
associated with "body"?

Or is this an frbr:CorporateBody ?

[7/5/13 6:48:45 PM] Tara Athan: Regarding "AuxiliaryParty - a entity in 
addition to the bearer of a deontic specification", isn't this a role? 
It is a different kind of role than the metadata <lrml:Role>, but 
nevertheless it is not an essential characteristic of an entity, but is 
a part that some entity plays (i.e. a role that it fills) relative to a 
particular deontic specification, for some time period that is a part, 
typically not the entirety, of its history.

Suggestion:

Bearer - a role of a deontic specification filled by the entity to which 
the specification is primarily directed
AuxiliaryParty - a role of a deontic specification filled by entities to 
which the specification is related, but not primarily directed
[7/5/13 7:12:06 PM] Tara Athan: "Compliance - a situation …"  - is there 
a neutral concept of "situation" that I could reference in the 
metamodel, or would this be better left as a subclass of "Thing"?
[7/5/13 7:46:29 PM] Tara Athan: Bearer - a role *in* a deontic 
specification filled by the entity to which the specification is 
primarily directed
AuxiliaryParty - a role *in* a deontic specification filled by entities 
to which the specification is related, but not primarily directed
[7/5/13 8:00:43 PM] Tara Athan: "LegalSource - Any source of legal norms 
represented in any format."
I don't understand this definition as written.

As we've constructed the syntax, Isn't the legal norm (playing the role 
of) a source, and if so, in relation to what (a LegalRuleML text)?

But this definition gives the impression that the legal norm *has* the 
source (from "source of legal norms" --> the source belongs to the legal 
norm, the legal norm has the source)

Some other suggestions, beyond what was in the Skype chat

FactualStatement - a statement that expresses a fact.
This wording helps our taxonomy - it tells us what FactualStatement is a 
subclass of - Statement. One question for clarification - is there any 
distinction between "legal" factual statements and those that do not 
have legal status? If a factual statement is formalized in LegalRuleML, 
can it be assumed to have legal status?

Regarding the glossary entry for lrml:Context, I believe we discussed 
earlier the idea that

{Text} + {Context} = {Rule}

However, this is not reflected in the Glossary definition.
Also the Glossary definition gives an impression, to me, that a context 
is only applicable to a single text.

Suggestion:

Original
Context - a context or legal interpretation that applies associations of 
the proper author, authorities, jurisdiction, source, and other 
characteristics to a rule and/or to parts of a rule.
-->
Context - a context or legal interpretation that may apply associations 
of the proper author, authorities, jurisdiction, source, and other 
characteristics to texts and/or to parts of texts, and through this 
application, generate rules.

Question: is there ever a case when a context is modular - that certain 
characteristics are specified in one block, other characteristics in 
another, and the rule is generated by the application of both? 
Alternately, can a context "import" another context, perhaps through the 
key/keyref mechanism?

[7/6/13 9:29:49 AM] Tara Athan: Thoughts on the relationship between 
entities in the LegalRuleML metamodel and entities in external legal 
ontologies:
When we mint a URL using a "key" attribute in a LegalRuleML text for, 
say, an "lrml:Obligation"

<lrml:Obligation key = "obl1">
  <ruleml:slot>
      <lrml:Bearer/>
      <ruleml:Var>Customer</ruleml:Var>
  </ruleml:slot>
   <ruleml:Implies> …</ruleml:Implies>
 </lrml:Obligation>
what sort of thing does that URL denote?

It's not referring to any particular obligation in the "real world" - 
such an obligation would require the binding of "<Var>Customer</Var>" to 
a particular person.
It appears to me that the URL denotes a formula about obligations - this 
also makes sense in a taxonomy (metamodel) where "Obligation", or better 
"lrmlmm:ObligationFormula" is a subclass of "lrmlmm:DeonticFormula'

Suggested definition for the metamodel:
ObligationFormula - The class of deontic formulas that express 
situations, acts, or courses of action to which a bearer is legally 
bound, and if it is not achieved or performed results in a violation.

Similar considerations apply to all the deontic formulas.

In the glossary definition of "Permission", we mention the two cases of 
weak and strong permission. Would it make sense to have an attribute or 
child of <lrml:Permission> to allow the specification of the kind of 
permission?
[7/6/13 10:59:22 AM] Tara Athan: Re: Reparation - a legal statement of 
the penalty that may apply if a rule is violated. A rule is violated 
where the condition of applicability (antecedent of the rule) holds and 
the deontic specification (consequent of the rule) has been violated.

There are a few things that are implicit in this definition. First, it's 
not true that for any rule, its consequent is a deontic specification 
(deontic formula?). So we are talking about only certain kinds of rules 
here. We don't yet have a LegalRuleML name for these rules ( 
prescriptive rule?). We do have schema type for them that uses this name 
(PrescriptiveRule.type.def).

Second, it seems interesting to me that the penalty is linked to the 
rule rather than the obligation. This means the penalty for violating an 
obligation may depend on the situation under which that obligation 
arose, not just the specification of the obligation itself. And this is 
why the child of a reparation must be a rule rather than a violation or 
obligation.
[7/6/13 11:11:02 AM] Tara Athan: More on LegalSource - if LegalSource is 
a role, then it is "filledBy" a LegalStatement, rather than being the 
"sameAs" the LegalStatement.
[7/6/13 12:42:03 PM] Tara Athan: Re "TemporalCharacteristic - legal 
qualifications and descriptions of the relationships between a temporal 
entity and legal elements to which the relationships can apply."

Looking at this definition now after having been away from it for a 
while and forgotten what we were thinking about at the time, I have to 
say that I haven't a clue what this says.

Breaking it into parts, the definition first says that a temporal 
characteristic is (one or a set of?) legal qualifications and 
descriptions. Does this mean legal descriptions or just descriptions? 
"And" is used - does it have to contain both qualifications and 
descriptions? What is a legal qualification?

This is a very vague definition, but as embodied in the syntax it is a 
much more limited concept.

In the syntax, it is an information object with a data structure of type 
"associative array" with one key for specifying a temporal entity, 
another for specifying a rule status, and a third for specifying a 
status development.
[7/6/13 1:07:44 PM] Tara Athan: The glossary does not contain a 
definition for "Association". In the xsd schemas we have two definitions

"an efficient description of a set of dyadic relationships of various types
        (source, strength, ...),  whereby each non-target entity in the 
association is related to
        each target in the association"

applied to Association.Node.choice, and

"a set of typed dyadic relationships from each non-target entity to each
            target"
applied to Association.Node.def
This is unusual in that for most other elements we have one definition 
associated with the .Node.def component, and no extra explanation with 
the Node.choice element.

I can't find a record of the discussion about this definition in the 
Skype chat.
[7/6/13 3:46:21 PM] Tara Athan: In the Glossary and meta-model, we have 
used the name "Role" for what is really only a subclass of roles - the 
agentive role. There are other roles in the metamodel, and it is 
difficult to make this clear.

Suggestion: In the metamodel, we rename lrmlmm:Role as 
lrmlmm:AgentiveRole. and we have a superClass of lrmlmm:Role, which has 
subclasses "AgentiveRole", LegalSourceRole", …
Other roles include
"lrmlmm:Authority" is a role played by a frbr:CorporateBody,
"lrmlmm:Jurisdiction" is a role played by a geographic region or 
subject-matter domain

Re: the various kinds of Statements (ConstitutiveStatement, etc) in our 
Glossary. There is a disconnect between the metamodel and the Glossary. 
The glossary gives a definition as if we were referring to a legal 
statement. But in the model "Statement" is a LegalRuleML statement, 
which is not a legal statement. (The url we use in

<lrml:ConstitutiveStatement key="cs1">...</lrml:...>

does not denote a legal statement.)

Suggestion:

LegalRuleML Statement - a class (in the sense of rdfs:Class) of RuleML 
Rules sharing a common template.

Template - a reification of an underspecified RuleML Defeasible Rule.

hasTemplate - a template of the LegalRuleML statement.

Tara