National Institute of Advanced Industrial Science and Technology 1 A Semantic Matchmaker for RDF/OWL- based Service Repositories Mirza Said National Institute.

11National Institute of Advanced Industrial Science and Technology

A Semantic Matchmaker for RDF/OWL-A Semantic Matchmaker for RDF/OWL-based Service Repositoriesbased Service Repositories

Mirza SaidMirza SaidNational Institute of Advanced Industrial Science National Institute of Advanced Industrial Science

and Technology (AIST)and Technology (AIST)Grid Technology Research CenterGrid Technology Research Center

JapanJapan

Building a Tool for OWL-S Service Building a Tool for OWL-S Service Discovery and CompositionDiscovery and Composition

22

National Institute of Advanced Industrial Science and Technology

OutlineOutline

IntroductionIntroduction

OWL-SOWL-S

Semantic service matchmaking and main Semantic service matchmaking and main issuesissues

Proposed matchmaking systemProposed matchmaking system

Summary and future workSummary and future work

33


BackgroundBackground

Many Web services are available in the WebMany Web services are available in the WebThe number is rapidly increasingThe number is rapidly increasing

Google, Yahoo, Amazon, B2B, and etc.Google, Yahoo, Amazon, B2B, and etc.

Main issue: automatic and accurate discovery of Main issue: automatic and accurate discovery of target web servicestarget web services

Semantic Interoperability is a major hurdle for Locating Services

Different terms are used for advertisements and requests

OWL-S: An OWL-S: An ontology of service conceptsontology of service concepts for for describing the properties and capabilities of Web describing the properties and capabilities of Web services in a machine interpretable formservices in a machine interpretable form

44


OWL-S: The Upper OntologyOWL-S: The Upper Ontology

Consists of four main classesConsists of four main classesServiceProfileServiceProfile: for advertising : for advertising and discovering servicesand discovering services

What organization provides the What organization provides the service (e.g., contact information)service (e.g., contact information)What function the service computes What function the service computes (e.g., inputs and outputs) (e.g., inputs and outputs) What characteristics the service has What characteristics the service has (e.g., service category)(e.g., service category)

ProcessModelProcessModel: a detailed description of a service's : a detailed description of a service's operationoperationServiceGroundingServiceGrounding: the details of how to access the : the details of how to access the serviceserviceServiceService: binds three classes via three object properties: binds three classes via three object properties

55


An OWL-S Service: A Car Selling ServiceAn OWL-S Service: A Car Selling Service

hasInput: hasInput: Familycar & YearFamilycar & Year

hasOutput: hasOutput: PricePrice

serviceProduct: serviceProduct: Used-vehicleUsed-vehicle

textDescription: textDescription: used car shop in Tokyoused car shop in Tokyo

Car-shophasInput

YearYearhasInput

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

PricePrice

hasOutputUNSPSC

Sales-activities

Used-vehicle

New-vehicle

serviceProduct

UNSPContology

““A used cars shop A used cars shop locating in Tokyo”locating in Tokyo”

textDescription

66


OWL-S Service MatchmakingOWL-S Service Matchmaking

servicerepository &matchmaker

providerregister

requesterquery

A provider A provider describes describes advertised services using an advertised services using an OWL-S compliant ontologyOWL-S compliant ontology

A requester A requester queriesqueries the repository using the same the repository using the same ontologyontology

Repository Repository matchesmatches the request and advertisements the request and advertisements and returns matched advertisements in relevance orderand returns matched advertisements in relevance order

hasInput: sedan & yearhasOutput: price

request hasInput: Familycar & yearhasOutput: priceserviceProduct: used-vehicletextDescription: used car shop in Tokyo

advertisement

77


Basic Principle of MatchingBasic Principle of Matching ((Paolucci et al. ISWC02)

An advertisement matches a An advertisement matches a request when request when

all request outputsall request outputs are matched are matched by advertisement outputs andby advertisement outputs andall advertisement inputsall advertisement inputs are are matched by request inputsmatched by request inputs

Guarantees that the Guarantees that the matched matched serviceservice provides all outputs provides all outputs requested by the requester, requested by the requester, and that and that the requesterthe requester provides all input required for provides all input required for correct operation to the correct operation to the matched servicematched service

FLEXIBLE MATCHby defining degree of match

between two classes/concepts

advertisement

inp

ut

ou

tpu

t

request

inp

ut

ou

tpu

t

88


Degree of MatchDegree of Match

providerrequesterservice

repository &matchmaker

registerquery

hasInput: sedan & yearhasOutput: price

hasInput: Familycar & yearhasOutput: priceserviceProduct: used-vehicletextDescription: used car shop in Tokyo

request

advertisement

inR: a request inputinA: an advertisement input degreeOfMatch(inA, inR) Paolucci et.al. ISWC02

if inR=inA then Exactif inA immediateSubclassOf inR then Exactif inA subclassOf inR then Plug-inif inR subclassOf inA then Subsumeotherwise Fail

Subsume

Exact

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

inR

inA

|Exact| > |Plug-in| > |Subsume| > |Fail|

99


Main IssuesMain Issues

Using ontology as a service requestUsing ontology as a service requestDifficult to composeDifficult to compose

The repository usually only support simple The repository usually only support simple queriesqueries

Each matchmaking system uses a different Each matchmaking system uses a different matching enginematching engine

Need a proprietary query engineNeed a proprietary query engine

Tighly-couple architectureTighly-couple architecture

1010


GoalGoal

Propose a matchmaking mechanism thatPropose a matchmaking mechanism thatEnables complex service matchingEnables complex service matching

e.g, optional pattern matching e.g, optional pattern matching Has a loosely-couple architectureHas a loosely-couple architecture

Matchmaker can switch from one RDF repository Matchmaker can switch from one RDF repository to another or send a simultaneous query to to another or send a simultaneous query to multiple repositoriesmultiple repositories

allows to use existing “standard” matching allows to use existing “standard” matching enginesengines

Not build and using another proprietary matching Not build and using another proprietary matching engineengine

Using SPARQL as the query language!!Using SPARQL as the query language!!Using SPARQL as the query language!!Using SPARQL as the query language!!

SPARQL pattern matchingSPARQL pattern matching

Repository + SPARQL query engineRepository + SPARQL query engine

Many free SPARQL query engines are available Many free SPARQL query engines are available

1111


SPARQL: Entailments MatchingSPARQL: Entailments Matching

?profile?profile ?in?inprofile:hasInput rdf:type

Query pattern:Query pattern:

RDF Graph:RDF Graph:

auto:Sedanauto:Sedan

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

auto:pauto:p auto:inauto:in

profile:hasInput rdf:type

auto:Hatchbackauto:Hatchback

Inferred RDF Graph:Inferred RDF Graph:

auto:pauto:p auto:inauto:inprofile:hasInput rdf:type



…

rdf:typerdf:type

uninferred graph pattern uninferred graph pattern (uGraph)(uGraph)

inferred graph pattern inferred graph pattern (iGraph): default graph(iGraph): default graph

SPARQL does not interpret RDF schema information or entailments

(treats RDF graph purely as data)

SPARQL does not interpret RDF schema information or entailments

(treats RDF graph purely as data)

1212


Basic ArchitectureBasic Architecture

Requester sends a SPARQL query to MatchmakerRequester sends a SPARQL query to MatchmakerMatchmaker Matchmaker rewritesrewrites the query and sends it to an the query and sends it to an OWL-S service (RDF) repositoryOWL-S service (RDF) repositoryMatchmaker Matchmaker ordersorders the results based on the the results based on the degree degree of match between classesof match between classes

queryrewriting SPARQL

queryengine

OWL-S/RDFdata

repository

Matchmaker RDF Repository System

query rewrittenquery

SPARQL query ServiceProviderServiceProviderserviceprovider

unorderedresults +

ordering info. result

orderingorderedresults

orderedresults

register

SPARQLinterface

query

clientinterface

servicerequester

1313


Why Rewriting?Why Rewriting?

queryrewriting SPARQL

queryengine

OWL-S/RDFdata

repository

Matchmaker RDF Repository System

query rewrittenquery

SPARQL query

query

ServiceProviderServiceProviderserviceprovider

unorderedresults +

ordering info. result

orderingorderedresults

orderedresults

register

SPARQLinterface

clientinterface

Need to order results based on class subsumption but they do

not contain the information necessary for result ordering

Results are

unordered

Rewrite the query s.t. the results will contain

the necessary info.

servicerequester

1414


Why Rewriting?Why Rewriting?Lack of Class Subsumption Info.Lack of Class Subsumption Info.

?profile?profile ?in?in

auto:pauto:p auto:inauto:in

Query solution:Query solution:




Inferred graph (iGraph):Inferred graph (iGraph):

…




rdf:typerdf:type

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

No class subsumption info. :Hatchback is an imm. subclass of Sedan

(prevent Exact & Plug-in match calc.)

1515


Why Rewriting?Why Rewriting?Cannot Match Superclass InstancesCannot Match Superclass Instances

Query solution: not existQuery solution: not exist



auto:Familycarauto:Familycar



auto:Carauto:Car

auto:Automobileauto:Automobilerdf:type

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

Cannot match superclass instances(prevent Subsume match calculation)

1616


Why Rewriting?Why Rewriting?Lack of property inclusiveness info.Lack of property inclusiveness info.



auto:Carauto:Car


auto:pauto:p auto:in1auto:in1profile:hasInput rdf:type

auto:Carauto:Car

auto:yearauto:yearrdf:type

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology

Do not know whether auto:p has only one input or more

(prevent Fail match calculation)

?profile?profile ?in?in

auto:pauto:p auto:in1auto:in1

Query solution:Query solution:

auto:in2auto:in2profile:hasInput

1717


Query Rewriting: Basic PrincipleQuery Rewriting: Basic Principle

Rewrite the requester query such that the Rewrite the requester query such that the matching results matching results will includewill include the lacking the lacking informationinformation

Class subsumptionClass subsumption

Superclass instancesSuperclass instances

Property inclusivenessProperty inclusiveness

1818


Query Rewriting: The RulesQuery Rewriting: The Rules

A requester query:A requester query:

Rewrite each type-matching pattern:Rewrite each type-matching pattern:

Type-matching pattern

a = rdf:type

Automobile

Car Truck

Sportcar Familycar

Sedan Minivan

Hatchback Two-doors

car-type ontology Car-Model

Toyota Nissan

Prius Camry Sentra QuestRAV4

Asia-car Europe-car

Honda

SELECT ?sWHERE ?s presents ?o.?o hasInputhasInput ?in1, ?in2. ?in1 a Familycar. ?in2 a Toyota.

car-model ontology

A graph pattern that matches k values of property p while restricting values to those from specific classes

1919


Get superclasses and subclasses of property value classes (rs = rdfs:subclassOf)Get instances whose values are from the

superclasses and subclassesGet property inclusiveness informationGet subsumption class information

2020


SELECT ?sWHERE ?s presents ?o.?o hasInput ?in1, ?in2. ?in1 a Familycar. ?in2 a Toyota.

Get superclasses and subclasses of Familycar and Toyota and binding them to variables (rs = rdfs:subclassOf)

Get hasInput instances whose values are from the superclasses and subclasses of Familycar and Toyota

Check whether hasInput is inclusiveness.i.e., whether it has only two values whose types are the superclass/subclass of Familycar and Toyota

Get the immediate superclasses and subclasses of Familycar and Toyota(rs = rdfs:subclassOf)

2121


Extracting Info. from Binding VariablesExtracting Info. from Binding VariablesIf ?otype1 is bound to some value then hasInput is a non-inclusive property If ?sup1 and ?isup1 are bound to some classthen the class is an immediate superclass of Familycar

If ?sub1 and ?isub1 are bound to some classthen the class is an immediate subclass of Familycar

2222


Service OrderingService Ordering

Service functions (service inputs & outputs)Service functions (service inputs & outputs)

Service characteristics (service category, Service characteristics (service category, products and etc.)products and etc.)

2323


Service Ordering (Service Function)Service Ordering (Service Function)

Given a requester query Given a requester query qq::?o?o hasInputhasInput ?in?in11 , …, , …, ?in?inmm . .

?in?in11 rdf:type rdf:type cincin11. … . … ?in?inmm rdf:type rdf:type cincinmm. .

A matched service instance A matched service instance ss::pp hasInputhasInput pinpin11 , …, , …, pinpinmm . .

pinpin11 rdf:type rdf:type sinsin11. … . … pinpinmm rdf:type rdf:type sinsinmm..

Degree of match for service inputs:Degree of match for service inputs:ddinin((qq, , ss)) = DIN(= DIN(cincin11, , sinsin11) + … + DIN() + … + DIN(cincinmm, , sinsinmm))

Note: if hasInput of Note: if hasInput of ss in non-inclusive then Fail in non-inclusive then Fail

Degree of match for service outputs:Degree of match for service outputs:ddoutout((qq, , ss)=DOUT()=DOUT(coutcout11, , soutsout11)+ … +DOUT()+ … +DOUT(coutcoutnn, , soutsoutnn))

Degree of match for service function (0 Degree of match for service function (0 ≤ ≤ d(d(qq, , ss) ) ≤≤ 1): 1):

d(d(qq, , ss)=)=nn xx ddinin((qq, , ss)) ++ mm xx ddoutout((qq, , ss))

2 x |Exact| x2 x |Exact| x mm xx nn

2424


Degree of Match CalculationDegree of Match Calculation

DIN(DIN(inRinR, , inAinA))

if if inRinR==inAinA return return ExactExact

if if inRinR immSuperclassOf immSuperclassOf inAinA

return return ExactExact

if if inRinR superclassOf superclassOf inAinA

return return Plug-inPlug-in

otherwise return otherwise return SubsumeSubsume

DOUT(DOUT(outRoutR, , outAoutA))

if if outRoutR==outAoutA return return ExactExact

if if outRoutR immSubclassOf immSubclassOf outAoutA

return return ExactExact

if if outRoutR subclassOf subclassOf outAoutA

return return Plug-inPlug-in

otherwise return otherwise return SubsumeSubsume

2525


Service Ordering (Service Characteristics)Service Ordering (Service Characteristics)

Given a requester query Given a requester query qq, where , where pp specifies a specifies a service characteristicservice characteristic

?o?o pp ?var?var11 , …, , …, ?var?varmm . .

?var?var11 rdf:type rdf:type cc11. … . … ?var?varmm rdf:type rdf:type ccmm. .

A matched service instance A matched service instance ss::pscpsc pp pcpc11 , …, , …, pcpcmm . .

pcpc11 rdf:type rdf:type scsc11. … . … pcpcmm rdf:type rdf:type scscmm..

Degree of match for a service characteristic:Degree of match for a service characteristic:ddpp((qq, , ss)) = DOUT(= DOUT(cc11, , scsc11) + … + DOUT() + … + DOUT(ccmm, , scscmm))

Degree of match for Degree of match for mm properties: properties:

2626


Overall ScoringOverall Scoring

The overall scoringThe overall scoring

dd((qq, , ss)=αx )=αx ddff((qq, , ss)) + (1-α) x + (1-α) x ddcc((qq, , ss)) , ,

where 0where 0≤≤αα≤≤11When When ddff((qq, , ss)) fails fails d(d(qq, , ss)) also fails also fails

Can be used to order general SPARQL query results Can be used to order general SPARQL query results by setting α=0by setting α=0

2727


Discussion and ConclusionsDiscussion and Conclusions

Research achievementsResearch achievementsUsing SPARQL as the query language for service Using SPARQL as the query language for service requesterrequester

Loosely-coupled architecture of the matchmaking Loosely-coupled architecture of the matchmaking systemsystemLighweight matchmakerLighweight matchmakerComplex service matching requestsComplex service matching requests

Major current challengeMajor current challengeHow to rewrite the requester query such that the How to rewrite the requester query such that the matching results will contain the information matching results will contain the information necessary for result ordering necessary for result ordering

2828


Discussion and ConclusionsDiscussion and Conclusions

Future workFuture workQuery rewriting rules for queries with SPARQL Query rewriting rules for queries with SPARQL UNION connecting type-matching patternsUNION connecting type-matching patterns

Degree of match based on the relationship Degree of match based on the relationship among classes and their property classes (e.g., among classes and their property classes (e.g., c2 is a property of c1)c2 is a property of c1)

Query rewriting rules for extracting the informationQuery rewriting rules for extracting the information

National Institute of Advanced Industrial Science and Technology 1 A Semantic Matchmaker for RDF/OWL- based Service Repositories Mirza Said National Institute.

Documents

service service

service hasinput

service computes

service request difficult

ontology of service

owls service discovery

query hasinput

ontology repository