This is the promised XML supplement to my 2012-02-22 email with subject line: [legalruleml] [ontology] Legal Domain Ontology.
About Monica’s remarks regarding Tara’s two points:
1. Extending Monica’s path name to a URL, I assume the required LKIF vocabulary will be available at http://www.oasis-open.org/apps/org/workgroup/legalruleml/ontology/lkif/consumerLaw (this URL could be easily changed, see 2.). Omitting the defeasibility attribute for the moment (to be discussed in connection with Guido’s yesterday’s draft), the current RuleML (specifically, Hornlogeq) can then be used for Monica’s example refining the premium rule (http://ruleml.org/papers/Primer) as shown below. Keep in mind that RuleML uses an ‘object-oriented’ or ‘striped-syntax’ XML, where object-like, Upper-cased ‘node’ elements alternate with property-like, lower-cased ‘edge’ elements. Some edge elements, such as <op> and <arg>, are optional because their omission does not produce ambiguity, but node elements cannot be ‘skipped’ in this way. To avoid confusion with XML’s general @id attribute (allowed on all elements), which assumes too restrictive types of values, a @node attribute with type xs:anyURI is used (allowed only on node elements). For simplicity (until we settle on the Temporal RuleML layer), I use an extra integer argument on the premium and spending relations for the time interval of their validity, where an integer such as 2012 represents an entire calendar year (interval) from the total order of AD years. A predecessor operation (which really should be applicable only to integers > 1) can then be used to get at what is just an individual constant <Ind>previous year</Ind> in the Primer. This is (preliminarily) implemented with RIF’s numeric-subtract built-in, keeping Monica’s assumption that the rule is used backward only, with the <Var>year</Var> parameter bound to a fixed integer. The premium rule’s <And> premise is assumed to be (left-to-right) sequential, first computing <Var>previous-year</Var> as numeric-subtract (an ‘active’, per-value function) of 1 from the premium <Var>year</Var>; then looking up a spending <Var>amount</Var> with a euro unit (represented as a ‘passive’, per-copy function) for the previous year; and finally ensuring that the amount was greater than or equal to 5000.
<Assert mapClosure="universal" mapDirection="backward" mapMaterial="no">
<Fun iri="http://www.w3.org/TR/rif-dtb/#func:numeric-subtract" per="value"/>
2. The above-illustrated RuleML syntax is independent from the ontology because vocabulary URLs such as …/lkif/consumerLaw#xyz would just need to be replaced with other ontologies’ vocabulary URLs such as …/superlegal/consumerLaw#xyz.
From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of monica.palmirani
Sent: Wednesday, February 15, 2012 9:01 PM
Subject: Re: [legalruleml] Legal Domain Ontology
good points! Thanks for this email.
1. I worked with the group of LKIF [http://www.estrellaproject.org/lkif-core/] in the Estrella project and I contributed in some parts [http://www.estrellaproject.org/doc/D1.1-LKIF-Specification.pdf; http://carneades.berlios.de/resources/Home/LKIF-Specification.pdf], so I am inclined to use it extensively for coping with one of the main goal of LegalRuleML: fill the gap between text and rule:
I want to use LKIF for different purposes:
a) to link the predicates/verbs of the rules to the connected ontology class. Example: Rel "premium" is a predicate so I want to connect this predicete with its definition in the ontology with the legal classe "premium". The same for "customer" considering the definition provided by the Consumer Law. This permits to link the predicate to the law definitions and to improve interoperability among different legal knowledge entgineers.
b) to link the text with the legal concepts. If the text said "end-user" instead of "customer" I can connect the Var "counterpart" to the legal concept of "customer" and in a such a way to favor the interoperability, the retrieval by concepts, the multilingual text, the interconnection among different base-rules.
c) to use the axiomes of the LKIF-core for the legal reasoning combined with the rules. The concept of "customer" in Consumer Law is a precise class of people with some properties. If I define these proprieties in the ontology, I can reuse it any time that I need it.
[Palmirani, et. al., Fill the Gap in the Legal Knowledge Modelling , RuleML2009]
<Assert mapClosure="universal" mapDirection="backward" mapMaterial="no">
<Implies timesBlock="#t2" ruleType="defeasible" id="rule1">
<Atom id="atm2" timesBlock="#t3">
<Rel>previous year spending </Rel>
<formula><Var>x</Var> >= 5000 euro</formula>
After this introduction, when I said " LegalRuleML wants to define the main legal classes of the rules components" I have in mind a different kind of ontology: not thematic ontology linked to a specific legal domain, but the anatomy of the "legal rule". The question is: which parameters or attributes make the "legal rule" special and interesting respect a normal rule? which parts we have to define for permitting a correct and powerfull legal reasoning ?
2. the syntax is independent to the ontology because I can point-out to different ontologies using "refersTo" attribute (not only to LKIF). My goal is different: to define the meta-ontology of the "legal rule". This work influences for sure the LegalRuleML tags definition. E.g. as naf was created for capturing a particular concept of negation, so we have to define, and after to represent with appropriate tags in LegalRuleML, the essence of the "legal rule". E.g. obligation: subject of the obligation, addresser, authority, level of obligation (civil, penal, administrative), jurisdiction of the obligation (only in Italy, in all the world, for all the facts or for only the fact happended in Italy), temporal validity of the obligation.
( 184.108.40.206 Semantic modeling in the LegalRuleML 1.0-core-spec http://www.oasis-open.org/apps/org/workgroup/legalruleml/download.php/44972/LegalRuleML0.1-core-spec-2012-01-30.doc)
Il 16/02/2012 00:37, Tara Athan ha scritto:
>From the discussion today it sounds like part of LegalRuleML will be an ontology of concepts in the legal domain.
Somethings I would like to see clarified are:
1. to what extent will this ontology be taken from existing ontologies, including upper ontologies, such as DOLCE , or legal domain ontologies such as lkif , and to what extent will it be new;
2. is it desirable that the parts of the syntax that are dependent on the ontology be separable (through modularization) from the parts that are independent.
I would argue that in #1 as much as possible be used from existing ontologies provided these ontologies have a broad acceptance, and for #2 as much separation as possible, so that if/when the ontology evolves or is replaced, it has minimal effects on the syntax that is independent of the legal domain.
--------------------------------------------------------------------- To unsubscribe, e-mail: firstname.lastname@example.org For additional commands, e-mail: email@example.com
Associate professor of Legal Informatics
School of Law
Alma Mater Studiorum Università di Bologna
Palazzo Dal Monte Gaudenzi - Via Galliera, 3
I - 40121 BOLOGNA (ITALY)
Tel +39 051 277217
Fax +39 051 260782