46
Enabling Reasoning with LegalRuleML Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield RuleML 2016, presented by Guido Governatori www.data61.csiro.au

Enabling Reasoning with LegalRuleML

Embed Size (px)

Citation preview

Enabling Reasoning withLegalRuleML

Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

RuleML 2016, presented by Guido Governatori

www.data61.csiro.au

Background

• Certain requirements needs to be fulfilled

I Isomorphism

I Reification (such asJurisdiction, Authority, andTemporal properties)

I Rule semantics

I Rule validity

I Defeasibility and Conflictsresolutions

I Normative effects (such asObligation, Permission,Prohibition, etc) and theirpersistence

I . . .

• Existing formalisms such as LKIF, ConDec, ContractLog,RuleML lack the support to fulfill such requirements

⇒LegalRuleML

2 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Background

• Certain requirements needs to be fulfilled

I Isomorphism

I Reification (such asJurisdiction, Authority, andTemporal properties)

I Rule semantics

I Rule validity

I Defeasibility and Conflictsresolutions

I Normative effects (such asObligation, Permission,Prohibition, etc) and theirpersistence

I . . .

• Existing formalisms such as LKIF, ConDec, ContractLog,RuleML lack the support to fulfill such requirements

⇒LegalRuleML

2 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Background

• Certain requirements needs to be fulfilled

I Isomorphism

I Reification (such asJurisdiction, Authority, andTemporal properties)

I Rule semantics

I Rule validity

I Defeasibility and Conflictsresolutions

I Normative effects (such asObligation, Permission,Prohibition, etc) and theirpersistence

I . . .

• Existing formalisms such as LKIF, ConDec, ContractLog,RuleML lack the support to fulfill such requirements⇒LegalRuleML

2 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

A Sample Contract

Section 1: (Policy of Price)

1.1 A “Premium Customer” is a customer who has spentmore than $10000 in goods. Premium Customers areentitled a 5% discount on new orders.

1.2 Goods marked as “Special Order” are subject to a 5%surcharge. Premium customers are exempt from specialorder surcharge.

1.3 . . . . . .

Section 2: (Purchase Order)

2.1 The purchaser shall follow the supplier price lists on thesupplier’s website.

2.2 The purchaser shall present supplier with a purchase orderfor the provision of goods within 7 days of thecommencement date.

2.3 . . . . . .

3 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML

• A rule interchange languageproposed by OASIS

• Extends RuleML with featuresspecific to legal domain

• Bridge the gap between naturallanguage description andsemantic norms

• Allows for modelling variouslaws, rules and regulations intomachine readable format

Association(s)

Context

Metadata

Statements

LegalRuleML Document

4 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML (cont.)

• Provides support toI model normative effectsI capture conflicting norms without making the set of rules

inconsistent.I prioritize norms in case both of them becomes applicableI norms that compensate the violation of other norms

• Similar to that of Defeasible Logic

5 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Defeasible Logic (DL)

• Rule-base, without disjunction

r : A(r) ↪→ C (r)where

I r is the unique identifier of the ruleI A(r) = a1, · · · , an the antecedent of the rule (where ai is a

literal)I ↪→= {→,⇒, } denotes the type of the rule (→=strict rule,⇒=defeasible rule, and =defeater)

I C (r) the consequent (or head) of the rule

• Classical negation is used in the heads and bodies of rulesI Negation-as-failure is NOT used but can be emulated

• Rules may support conflicting conclusions• Direct Skeptical: conflicting rules do not fire

I consistency is preserved

• Priorities on rules (superiority relation) may be used to resolvesome conflicts among rules

6 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Defeasible Logic (cont.)

• Constructive proof theory

• Flexible and computationally efficientI linear w.r.t. the number of literals

• Many extensions and applicationsI policy-based intentionI e-contracts analysis and monitoringI web service compositionI . . .

7 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Modal Defeasible Logic (MDL)

• Extension of DL with the support of modal operators

• The language consists of:I a set PROP of propositional atomsI Lit = PROP ∪ {¬p | p ∈ PROP } represents a set of literalsI If q is literal (∼q is its complement)

– if q is a positive literal p, then ∼q is ¬p– if q is ¬p, then ∼q is p

I MOD denotes a set of model operators

• Introduced the concept of contrary-to-duty obligation in thehead of the ruleFor an expression a⊗ b, the intuitive reading is that

I if a is possible, then a is the first choice and b is the secondone;

I if ¬a holds, i.e., a is violated, then b is the actual choice.

8 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Modal Defeasible Logic (MDL)

• Extension of DL with the support of modal operators

• The language consists of:I a set PROP of propositional atomsI Lit = PROP ∪ {¬p | p ∈ PROP } represents a set of literalsI If q is literal (∼q is its complement)

– if q is a positive literal p, then ∼q is ¬p– if q is ¬p, then ∼q is p

I MOD denotes a set of model operators

• Introduced the concept of contrary-to-duty obligation in thehead of the ruleFor an expression a⊗ b, the intuitive reading is that

I if a is possible, then a is the first choice and b is the secondone;

I if ¬a holds, i.e., a is violated, then b is the actual choice.

8 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Modal Defeasible Logic (MDL)

• Extension of DL with the support of modal operators

• The language consists of:I a set PROP of propositional atomsI Lit = PROP ∪ {¬p | p ∈ PROP } represents a set of literalsI If q is literal (∼q is its complement)

– if q is a positive literal p, then ∼q is ¬p– if q is ¬p, then ∼q is p

I MOD denotes a set of model operators

• Introduced the concept of contrary-to-duty obligation in thehead of the ruleFor an expression a⊗ b, the intuitive reading is that

I if a is possible, then a is the first choice and b is the secondone;

I if ¬a holds, i.e., a is violated, then b is the actual choice.

8 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• LegalRuleML is inherited from RuleMLI i.e., elements related to RuleML can be done using the

mapping provided in (Governatori, 2005)

• For instance, the following LegalRuleML code can betransformed into the literal pay(Purchaser ,Supplier)

1 <ruleml:Atom key=":atom109">2 <ruleml:Rel iri="pay"/>3 <ruleml:Ind>Purchaser</ruleml:Ind>4 <ruleml:Ind>Supplier</ruleml:Ind>5 </ruleml:Atom>

9 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• LegalRuleML is inherited from RuleMLI i.e., elements related to RuleML can be done using the

mapping provided in (Governatori, 2005)

• For instance, the following LegalRuleML code can betransformed into the literal pay(Purchaser ,Supplier)

1 <ruleml:Atom key=":atom109">2 <ruleml:Rel iri="pay"/>3 <ruleml:Ind>Purchaser</ruleml:Ind>4 <ruleml:Ind>Supplier</ruleml:Ind>5 </ruleml:Atom>

9 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

Type of Statements in LegalRuleML

StatementsNormStatements

ConstitutiveStatements

PrescriptiveStatements

Violation-ReparationStatements

ReparationStatements

PenaltyStatements

FactualStatement

OverrideStatement

which correspond to DL elements

LegalRuleML DL

Factual Statement FactConstitutive Statement Strict rulePrescriptive Statement Defeasible ruleOverride Statement Superiority relation

Violation-Reparation Statement describe later

10 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

Type of Statements in LegalRuleML

StatementsNormStatements

ConstitutiveStatements

PrescriptiveStatements

Violation-ReparationStatements

ReparationStatements

PenaltyStatements

FactualStatement

OverrideStatement

which correspond to DL elements

LegalRuleML DL

Factual Statement FactConstitutive Statement Strict rulePrescriptive Statement Defeasible ruleOverride Statement Superiority relation

Violation-Reparation Statement describe later

10 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1

rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

An Example

1 <lrml:PrescriptievStatement key="r1">2 <ruleml:Rule key=":ruletemplate1">3 <ruleml:if>4 <ruleml:And>5 <ruleml:Atom key=":atom2">6 <ruleml:Rel iri=":specialOrder"/>7 <ruleml:Ind>X</ruleml:Ind>8 </ruleml:Atom>9 </ruleml:And>

10 </ruleml:if>11 <ruleml:then>12 <lrml:Obligation>13 <ruleml:Atom key=":atom3">14 <ruleml:Rel iri=":surcharge"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:then>19 </ruleml:Rule>20 </lrml:PrescriptievStatement>

rule label r1rule typedefeasible rule

rule body specialOrder(X )

rule head OBLsurcharge(X )

r1 : specialOrder(X )⇒ OBLsurcharge(X )

11 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

Some notes:

• Penalty statements is essentially a list of deontic literals thatcan be appended to the head of the rule(s) concerned using⊗-expression based on the key referenced by thecorresponding Reparation statements.

• The same penalty statement may applies to many differentprescriptive statements, as described in the Reparationstatements.

12 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Statements

Some notes:

• Penalty statements is essentially a list of deontic literals thatcan be appended to the head of the rule(s) concerned using⊗-expression based on the key referenced by thecorresponding Reparation statements.

• The same penalty statement may applies to many differentprescriptive statements, as described in the Reparationstatements.

12 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Other constructs

However,

• Some elements in LegalRuleML are not that intuitive

• For instance, LegalRuleML provides two elements todetermine whether an obligation or a prohibition of an objecthas been fulfilled (<lrml:Compliance>) or violated(<lrml:Violation>)

Definition

• A compliance is an indication that an obligation has beenfulfilled or a prohibition has not been violated.

• A violation is an indication that an obligation or prohibitionhas been violated.

13 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Compliance and Violation

1 <lrml:PrescriptiveStatement key="ps2">2 <ruleml:Rule key=":ruletemplate2">3 <ruleml:if>4 <ruleml:And key=":and1">5 <lrml:Violation keyref="#item3"/>6 <lrml:Permission>7 <ruleml:Atom key=":atom4">8 <ruleml:Rel iri=":rel1"/>9 <ruleml:Ind>X</ruleml:Ind>

10 </ruleml:Atom>11 </lrml:Permission>12 <lrml:Obligation key="oblig1">13 <ruleml:Atom key=":atom5">14 <ruleml:Rel iri=":rel2"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:And>19 </ruleml:if>20 . . .

• need to determine whetherthe object with key=item3has been violated or not

• Two cases: when item3 is

(i) a literal

(ii) a rule

14 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Compliance and Violation

1 <lrml:PrescriptiveStatement key="ps2">2 <ruleml:Rule key=":ruletemplate2">3 <ruleml:if>4 <ruleml:And key=":and1">5 <lrml:Violation keyref="#item3"/>6 <lrml:Permission>7 <ruleml:Atom key=":atom4">8 <ruleml:Rel iri=":rel1"/>9 <ruleml:Ind>X</ruleml:Ind>

10 </ruleml:Atom>11 </lrml:Permission>12 <lrml:Obligation key="oblig1">13 <ruleml:Atom key=":atom5">14 <ruleml:Rel iri=":rel2"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:And>19 </ruleml:if>20 . . .

• need to determine whetherthe object with key=item3has been violated or not

• Two cases: when item3 is

(i) a literal

(ii) a rule

14 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Compliance and Violation

1 <lrml:PrescriptiveStatement key="ps2">2 <ruleml:Rule key=":ruletemplate2">3 <ruleml:if>4 <ruleml:And key=":and1">5 <lrml:Violation keyref="#item3"/>6 <lrml:Permission>7 <ruleml:Atom key=":atom4">8 <ruleml:Rel iri=":rel1"/>9 <ruleml:Ind>X</ruleml:Ind>

10 </ruleml:Atom>11 </lrml:Permission>12 <lrml:Obligation key="oblig1">13 <ruleml:Atom key=":atom5">14 <ruleml:Rel iri=":rel2"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:And>19 </ruleml:if>20 . . .

• need to determine whetherthe object with key=item3has been violated or not

• Two cases: when item3 is

(i) a literal

(ii) a rule

14 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping– Compliance and Violation

1 <lrml:PrescriptiveStatement key="ps2">2 <ruleml:Rule key=":ruletemplate2">3 <ruleml:if>4 <ruleml:And key=":and1">5 <lrml:Violation keyref="#item3"/>6 <lrml:Permission>7 <ruleml:Atom key=":atom4">8 <ruleml:Rel iri=":rel1"/>9 <ruleml:Ind>X</ruleml:Ind>

10 </ruleml:Atom>11 </lrml:Permission>12 <lrml:Obligation key="oblig1">13 <ruleml:Atom key=":atom5">14 <ruleml:Rel iri=":rel2"/>15 <ruleml:Ind>X</ruleml:Ind>16 </ruleml:Atom>17 </lrml:Obligation>18 </ruleml:And>19 </ruleml:if>20 . . .

• need to determine whetherthe object with key=item3has been violated or not

• Two cases: when item3 is

(i) a literal

(ii) a rule

14 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL MappingCase 1 [Element as literal] Assume that p is elementthat item3 refers to, from the code above we have:

ps2 : PERrel1,OBLrel2, violate(p)⇒ C (ps2)

Some observations:• The case where item3 is a literal is simple. It can be

appended to the body as a precondition to activate the ruleusing the table below:

q OBLq FORq

Compliance q OBLq, q FORq,¬qViolation ¬q OBLq,¬q FORq, q

• i.e., if p = OBLq, then the above rule becomes:

ps2 : PERrel1,OBLrel2,OBLq,¬q ⇒ C (ps2)

15 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL MappingCase 1 [Element as literal] Assume that p is elementthat item3 refers to, from the code above we have:

ps2 : PERrel1,OBLrel2, violate(p)⇒ C (ps2)

Some observations:• The case where item3 is a literal is simple. It can be

appended to the body as a precondition to activate the ruleusing the table below:

q OBLq FORq

Compliance q OBLq, q FORq,¬qViolation ¬q OBLq,¬q FORq, q

• i.e., if p = OBLq, then the above rule becomes:

ps2 : PERrel1,OBLrel2,OBLq,¬q ⇒ C (ps2)

15 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL MappingCase 1 [Element as literal] Assume that p is elementthat item3 refers to, from the code above we have:

ps2 : PERrel1,OBLrel2, violate(p)⇒ C (ps2)

Some observations:• The case where item3 is a literal is simple. It can be

appended to the body as a precondition to activate the ruleusing the table below:

q OBLq FORq

Compliance q OBLq, q FORq,¬qViolation ¬q OBLq,¬q FORq, q

• i.e., if p = OBLq, then the above rule becomes:

ps2 : PERrel1,OBLrel2,OBLq,¬q ⇒ C (ps2)

15 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

Case 2 [Element as Rule] In case of violation, we need to verifywhether referenced rule is either:

• inapplicable such that there a non-provable literal in theantecedent of the rule

• the immediate consequents of the rule is defeated or overruledby a conflicting conclusion.

16 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

Case 2 [Element as Rule] In case of violation, we need to verifywhether referenced rule is either:

• inapplicable such that there a non-provable literal in theantecedent of the rule

• the immediate consequents of the rule is defeated or overruledby a conflicting conclusion.

16 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

Definition (Rule Status)

Let D = (F ,R, >) be a defeasible theory, let Σ be the language ofD. For every r ∈ Rb, rc denotes the rule referenced by the elementWe define verifyBody(D) = (F ,R ′, >′) where:

R ′ = R \ Rd ∪ { r+c : A(rc)⇒ inf (rc),−c : ⇒ ¬inf (rc),

r−cv : ¬inf (rc)⇒ violation(rc),

r+cc : inf (rc), comply(C (rc , 1))⇒ compliance(rc),

r+cv : inf (rc), violate(C (rc , 1))⇒ violation(rc),

>′=> ∪{ r+c > r−c }

17 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• For each rc , inf (rc), ¬inf (rc), compliance(rc) andviolation(rc) are new atoms not in the language of thedefeasible theory.

• inf (rc) and ¬inf (rc) determine whether a rule is in force.

• If rc is in force, then we verify whether:

I the first literal that appears at the head of rc is compliant i.e.,compliance(rc) or

I rc is violated i.e., violated(rc)

18 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• For each rc , inf (rc), ¬inf (rc), compliance(rc) andviolation(rc) are new atoms not in the language of thedefeasible theory.

• inf (rc) and ¬inf (rc) determine whether a rule is in force.

• If rc is in force, then we verify whether:

I the first literal that appears at the head of rc is compliant i.e.,compliance(rc) or

I rc is violated i.e., violated(rc)

18 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• For each rc , inf (rc), ¬inf (rc), compliance(rc) andviolation(rc) are new atoms not in the language of thedefeasible theory.

• inf (rc) and ¬inf (rc) determine whether a rule is in force.

• If rc is in force, then we verify whether:

I the first literal that appears at the head of rc is compliant i.e.,compliance(rc) or

I rc is violated i.e., violated(rc)

18 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• For each rc , inf (rc), ¬inf (rc), compliance(rc) andviolation(rc) are new atoms not in the language of thedefeasible theory.

• inf (rc) and ¬inf (rc) determine whether a rule is in force.

• If rc is in force, then we verify whether:I the first literal that appears at the head of rc is compliant i.e.,

compliance(rc) or

I rc is violated i.e., violated(rc)

18 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

LegalRuleML to MDL Mapping

• For each rc , inf (rc), ¬inf (rc), compliance(rc) andviolation(rc) are new atoms not in the language of thedefeasible theory.

• inf (rc) and ¬inf (rc) determine whether a rule is in force.

• If rc is in force, then we verify whether:I the first literal that appears at the head of rc is compliant i.e.,

compliance(rc) orI rc is violated i.e., violated(rc)

18 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Conclusions

• We proposed a mapping of legal norms represented usingLegalRuleML into DL

• With current LegalRuleML specifications, strange resultsmight appear if an element <lrml:violation> (or<lrml:compliance>) appears at the head of astatement—additional restrictions are required.

• The real impedance is the lack of dedicated and reliableinfrastructure

• As future work, we are planning to incorporate the presentedapproach into some smart-contract enabled systems

19 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Conclusions

• We proposed a mapping of legal norms represented usingLegalRuleML into DL

• With current LegalRuleML specifications, strange resultsmight appear if an element <lrml:violation> (or<lrml:compliance>) appears at the head of astatement—additional restrictions are required.

• The real impedance is the lack of dedicated and reliableinfrastructure

• As future work, we are planning to incorporate the presentedapproach into some smart-contract enabled systems

19 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Conclusions

• We proposed a mapping of legal norms represented usingLegalRuleML into DL

• With current LegalRuleML specifications, strange resultsmight appear if an element <lrml:violation> (or<lrml:compliance>) appears at the head of astatement—additional restrictions are required.

• The real impedance is the lack of dedicated and reliableinfrastructure

• As future work, we are planning to incorporate the presentedapproach into some smart-contract enabled systems

19 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Conclusions

• We proposed a mapping of legal norms represented usingLegalRuleML into DL

• With current LegalRuleML specifications, strange resultsmight appear if an element <lrml:violation> (or<lrml:compliance>) appears at the head of astatement—additional restrictions are required.

• The real impedance is the lack of dedicated and reliableinfrastructure

• As future work, we are planning to incorporate the presentedapproach into some smart-contract enabled systems

19 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

Questions?

20 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield

References

Antoniou, Grigoris et al. (2001). “Representation Results for Defeasible Logic”.In: ACM Transactions on Computational Logic 2.2, pp. 255–286.

Gordon, Thomas F., Guido Governatori and Antonino Rotolo (2009). “Rules andNorms: Requirements for Rule Interchange Languages in the Legal Domain”.In: RuleML’09. Springer Heidelberg, pp. 282–296.

Governatori, Guido (2005). “Representing Business Contracts in RuleML”. In:International Journal of Cooperative Information Systems 14.2–3, pp. 181–216.

Nute, Donald (2001). “Defeasible logic: Theory, Implementation andApplications”. In: Proceedings of the 14th International Conference onApplications of Prolog (INAP 2001). Springer, Berlin, pp. 151–169.

21 | Enabling Reasoning with LegalRuleML | Ho-Pun Lam, Mustafa Hashmi, and Brendan Scofield