Experiences Using Deliberation RuleML 1.01 as a Rule Interchange Language

Experiences Using Deliberation RuleML 1.01 asa Rule Interchange Language

Do Rulebases Need an External Vocabulary?

The 8th International Web Rule Symposium Prague, Czech Republic, August 18-20, 2014

Matthias Tylkowski and Martin MüllerBinarypark, Erich-Weinertstr. 1, 03044 Cottbus, Germany

RuleML Basics

• Individuals, Data, Atomic statements, logical

formulas, rules, quantifiers

• Hierarchical sub-languages – RuleML lattice

• Support for various semantics

• Schema configurability via MYNG 1.01

• Design based on Relax NG and XML Schema

The 8th International Web Rule Symposium Prague, Czech Republic, August 18-20, 2014 2

http://wiki.ruleml.org/index.php/Language_Lattice

http://deliberation.ruleml.org/1.01/myng/

UServ Product Derby

• Describes a virtual car insurance company

• Business model based on business rules

• Traditional object-oriented data model

o classes (Car, Driver, ConvertibleCar, …),

o properties (price, eligibilityScore, potentialTheftRating )

o objects (car1, car2, …)


If all of the following are true, then the car’s potential theft rating is low:- car’s price is less that $20,000- car model is not on the list of “High Theft Probability Auto”

Any senior driver is not a young driver.

3

Individuals, Data and Variables (1)


RuleML Support:

• IRIs (@iri) to provide unique names for individuals. Typing

of individuals (@type)

• Declaration of property values for individuals.

<Atom>

<oid>

<Ind iri="http://userv.org/ontology/i/a9fccd3a-a851-4e80-de3a8c2e15ba"

type="http://userv.org/ontology/LuxuryCar">

</Ind>

</oid>

<slot>

<Rel iri="http://userv.org/ontology/price">price</Rel>

<Data xsi:type="xs:positiveInteger">39000</Data>

</slot>

</Atom>

4



<Assert>

<Atom>

<Rel iri="http://userv.org/ontology/carModel"/>


type="http://userv.org/ontology/LuxuryCar">car4</Ind>

<Data xsi:type="xs:string">Honda Odyssey</Data>

</Atom>

<Atom>

<Rel iri="http://userv.org/ontology/price">price</Rel>




</Atom>

</Assert>

RuleML Support:

• Best practice of declarative definition of instances (facts)

by atoms/formulas (𝑙𝑢𝑥𝑢𝑟𝑦𝐶𝑎𝑟 becomes a type)

𝑙𝑢𝑥𝑢𝑟𝑦𝐶𝑎𝑟 𝑐𝑎𝑟4 ∧ 𝑐𝑎𝑟𝑀𝑜𝑑𝑒𝑙 𝑐𝑎𝑟4, ′Honda Odyssey′ ∧ 𝑝𝑟𝑖𝑐𝑒 𝑐𝑎𝑟4, 39000 .

5




<Var type="http://userv.org/ontology/ConvertibleCar">C</Var>

RuleML Support:

• Beyond the logic programming tradition, allows typing of XML data(@xsi:type). However, only RuleML supported XML datatypes are permitted.

• Typed variables (@type).

<Data type="us:Rating">

<us:ratingValue xsi:type="xs:positiveInteger">4</us:ratingValue>

<us:ratingVerbalization xml:lang="en">

high

</us:ratingVerbalization>

</Data>

6

• No typing of User-defined structured data

UServ - Design


• Object instances (such as specific cars) map to

RuleML individuals

• Classes map to RuleML unary relations.

• Type classification (aka classes) can also be

declared explicitly inside typed variables or typed

individuals

• Properties map to RuleML binary relations

• Rules are based on ontology models

• UServ rules are based on relations and properties

defined by an external ontology (accessible via RuleML @type ) – the rule vocabulary .

7

Encoding Types/Classes

<Atom>

<Rel iri="us:Car">car</Rel>

<Var>C</Var>

</Atom>


RuleML Support:

• Type/class encoding using variables or individuals

• Type/class encoding using atoms with unary relations

<Ind type="us:Car">car4</Ind>

<Var type="us:Car">C</Var>

∀𝐶 𝑐𝑎𝑟(𝐶)

8

𝑐𝑎𝑟(𝑐𝑎𝑟4)

Multi-typing

RuleML Support:

• Explicit typing of individuals


<Assert>

<Atom>

<Rel iri="http://userv.org/ontology/Car">car</Rel>

<Ind iri="http://userv.org/ontology/i/a9fccd3a-a851-4e80">car1</Ind>

</Atom>

<Atom>

<Rel iri="http://userv.org/ontology/ConvertibleCar">car</Rel>


</Atom></Assert>

𝑐𝑎𝑟 𝑐𝑎𝑟1 ∧ 𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒𝐶𝑎𝑟(𝑐𝑎𝑟1)

9

Inheritance (1)

RuleML Support:

• No internal subClassOf inheritance support

• Define inheritance externally (e.g. RDFS) or via of rules

<Implies>

<Atom>

<Rel iri="us:ConvertibleCar">convertibleCar</Rel>

<Var>X</Var>

</Atom>

<Atom>

<Rel iri="us:Car">car</Rel>

<Var>X</Var>

</Atom>

</Implies>


∀𝐶 𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒𝐶𝑎𝑟 𝐶 → 𝑐𝑎𝑟(𝐶)

10

Inheritance (2)


<owl:Class rdf:about="http://userv.org/ontology/ConvertibleCar">

<rdfs:subClassOf rdf:resource="http://userv.org/ontology/Car"/></owl:Class>

<Atom>

<Rel iri="rdfs:subClassOf">subClassOf</Rel>

<arg index="1">

<Rel iri="us:ConvertibleCar" type="owl:Class">convertibleCar</Rel>

</arg>

<arg index="2">

<Plex>

<Rel iri="us:Car" type="owl:Class">car</Rel>

<Rel iri="us:LuxuryCar" type="owl:Class">luxuryCar</Rel>

</Plex>

</arg></Atom>

RuleML Support:• Rule-based inheritance may introduce complexity in execution.

• Usually search (backtracking) used for multiple inheritance

• Keep inheritance in external vocabulary or use higher-order syntax

𝐶𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒𝐶𝑎𝑟 ⊑ 𝐶𝑎𝑟

∀𝐶 ([𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒𝐶𝑎𝑟 𝐶 → 𝑐𝑎𝑟 𝐶 ] ∨ [𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒𝐶𝑎𝑟 𝐶 → 𝑙𝑢𝑥𝑢𝑟𝑦𝐶𝑎𝑟 𝐶 ]).

11

Encoding Properties (1)


<Atom>

<Rel iri="us:price">price</Rel>







</Atom>

<Atom>





</Atom>

∀𝐶 𝑐𝑎𝑟(𝐶) ∧ 𝑝𝑟𝑖𝑐𝑒 𝐶, 25000 .

luxuryCar(𝑐𝑎𝑟4) ∧ 𝑝𝑟𝑖𝑐𝑒 𝑐𝑎𝑟4, 39000 .

12

Encoding Properties(2)


<Atom>

<Rel iri="us:airbag">airbags</Rel>

<Var type="http://userv.org/ontology/Car">C</Var>

<Plex>

<Data xsi:type="xs:string">driverAirbag</Data>

<Data xsi:type="xs:string">frontAirbag</Data>

</Plex>

</Atom>

<Atom>

<Rel iri="us:theftRating">theftRating</Rel>

<Var type="http://userv.org/ontology/LuxuryCar">C</Var>

<Data type="us:Rating" >

<us:ratingValue xsi:type="xs:positiveInteger">4</us:ratingValue>

<us:ratingVerbalization xsi:type="xs:string" xml:lang="en">

high

</us:ratingVerbalization>

</Data>

</Atom>

∀𝐶 𝑙𝑢𝑥𝑢𝑟𝑦𝐶𝑎𝑟 𝐶 ∧ 𝑡ℎ𝑒𝑓𝑡𝑅𝑎𝑡𝑖𝑛𝑔 𝐶, 𝑟𝑎𝑡𝑖𝑛𝑔(4, ′high′) .

13

∀𝐶 𝑐𝑎𝑟 𝐶 ∧ 𝑎𝑖𝑟𝑏𝑎𝑔𝑠 𝐶, [′driverAirbag′, ′𝑓𝑟𝑜𝑛𝑡𝐴𝑖𝑟𝑏𝑎𝑔′]

Encoding Properties (3)• Definition of properties via external ontology


<Atom>



<Plex>

<Data xsi:type="xs:string">high</Data>


</Plex>

</Atom><owl:DatatypeProperty

rdf:about="http://userv.org/ontology/price">

<rdfs:domain rdf:resource=http://userv.org/ontology/Car/>

<rdfs:range>

<owl:Class>

<owl:unionOf rdf:parseType="Collection">

<owl:Class rdf:about="xs:positiveInteger"/>

<owl:Class rdf:about="xs:string"/>

</owl:unionOf>

</owl:Class>

</rdfs:range>

</owl:DatatypeProperty>

14

http://userv.org/ontology/Car

Built-ins

RuleML Support:

• Tests such as for numerical constraints and list memberships

• Operators with an @iri attribute refer to libraries of built-in

predicates such as one defined by RIF


<Atom>

<Rel iri="pred:numeric-less-than-or-equal">gt</Rel>

<Var type="xs:positiveInteger">P</Var>

<Data xsi:type="xs:integer">45000</Data>

</Atom>

𝑃 ≤ 45000.

15

http://www.w3.org/TR/rif-dtb/

Collection membership (1)

RuleML Support:

• List membership tests via use of collections (<Plex>) in atoms


<Atom>

<Rel iri="prolog:member">member</Rel>

<Var>M</Var>

<Plex>


<Ind iri="http://userv.org/ontology/i/5bff7cc6-72f7-4be3">car3</Ind>

<Ind iri="http://userv.org/ontology/i/5eb91389-f1f3-473c">car5</Ind>

</Plex>

</Atom>

16

Collection membership (2)

RuleML Support:

• Collections must be repeatedly given wherever needed

• Best practice to define collection constants via equality


<Equal oriented="yes">

<Ind>HighTheftProbabilityAutoList</Ind>

<Plex>


<Ind iri="http://userv.org/ontology/i/5bff7cc6-72f7-4be3">car3</Ind>

<Ind iri="http://userv.org/ontology/i/5eb91389-f1f3-473c">car5</Ind>

</Plex>

</Equal>

17

Rules and Rulebases (1)

RuleML Support:

• Rules are defined similarly with logical implications, but

many flavours are supported. A rulebase is a set of rules.


<Implies direction="backward">

<Naf>

<Atom>

<Rel iri="http://userv.org/ontology/hasRollBar">hasRollBar</Rel>

<Var type="http://userv.org/ontology/ConvertibleCar">C</Var>

</Atom>

</Naf>

<Atom>

<Rel iri="http://userv.org/ontology/injuryRating">injuryRating</Rel>

<Var>C</Var>

<Data><us:Value xsi:type="xs:positiveInteger">5</us:Value></Data>

</Atom>

</Implies>

If the car is a convertible and has no roll

bar, then the potential occupant injury is

extremely high.

18

Rules and Rulebases (2)RuleML Support:

• Limited support for annotations of rulebases: o Logic programming defines a relation by rules. All rules referring

that relation are grouped

o Production rules define a rulebase towards performing a specific task

• No global identity for rulebases (@iri)


<Rulebase xml:id="theftRating">



<meta>

<Atom>

<Rel iri="http://userv.org/ontology/theftRating">theftRating</Rel>

</Atom>

</meta>

…

</Rulebase>

19

Rulebase Metrics

• This rulebase covers around 40% of the use case

• We encoded:

o 30 rules

o 32 predicates, 5 built-ins

o 11 compound facts, 5 descriptions of individuals

o 7 usages of Naf

o 2 usages of Neg


Conclusion


• RuleML 1.01 can be used as a rule interchange language

• Improvements on handling types, inheritance, collections

and global constants are possible

• UServ design used classes and properties, defined by the

UServ ontology

• Our knowledge base uses NafNegHornlogEq family

• However, except for collections we do not use functions

("Datalog with Plex")

• Equality is used only to define collections as individuals

• Strong negation is used twice because of an explicit

negative information

• NafDatalog+FiniteCollections would be great

21

http://deliberation.ruleml.org/1.01/xsd/nafneghornlogeq.xsd

Acknowledgments

We would like to thank to Tara Athan, Harold Boley,

Adrian Giurca and Adrian Paschke for their essential

feedback and insights.


Useful information sources

• Tutorial introduction (http://ruleml.org/papers/Primer)

• RuleML MediaWiki (http://wiki.ruleml.org)

• RuleML Blog & Social Mediazine

(http://blog.ruleml.org)

• Mailing lists

(http://wiki.ruleml.org/index.php/Mailing_Lists)

• Technical Groups

(http://wiki.ruleml.org/index.php/Technical_Groups)

• RuleML sources hosted on Github

(https://github.com/RuleML)


http://ruleml.org/papers/Primer/RuleMLPrimer2012-08-09/

http://wiki.ruleml.org/

http://blog.ruleml.org/

http://wiki.ruleml.org/index.php/Mailing_Lists

http://wiki.ruleml.org/index.php/Technical_Groups

https://github.com/RuleML

Experiences Using Deliberation RuleML 1.01 as a Rule Interchange Language

Presentations & Public Speaking

czech republic

deliberation ruleml

rule vocabulary

ruleml individualsclasses

ruleml binary relationsrules

ruleml unary relations

typing of xml data

data model classes car