Ontology Networks

Document Sample
Ontology Networks Powered By Docstoc
					Querying Complex Information

             Doctoral School
             April 2007
             Fabio Porto
                                           Fábio Porto


   This course compiles the work I‟ve being
    developing on modeling and querying
    ontology-based applications;
                                     Fábio Porto


   WSMO,WSML Overview
   Semantic Web Service Discovery
   Querying multiple ontologies
QoS-enabled Semantic Web Service
                                                          Fábio Porto


   To develop a semantic approach for finding
    published web services based on high level
    user queries (goals)
    –   Includes:
            Semantic description of web services
            Model for quality of services
            Matchmaking (between goals and WS descriptions)
            Ranking
                                           Fábio Porto


   Project developed as part of the DIP FP6
    Integrated Project supported by the
    European union;
   DIP‟s objective was to develop and extend
    Semantic Web and Web Service
    technologies in order to produce a new
    technology infrastructure for Semantic Web
    Services (SWS)
   Project ran from 2004 to 2006
                                               Fábio Porto

Project outcomes

   WSMO model
   WSML language in different variants
   WSMX execution environment
   Tools: editing, repositories,discovery,…
                               Fábio Porto

1.   WSMO Overview
2.   Discovery in DIP
3.   Introduction on QoS-enabled
4.   QoS modelling
5.   QoS-based matchmaking
6.   Conclusion
         1. WSMO Overview

Adapted from - WSMO Tutorial
Michael Stollberg, Titi Roman, Holger Lausen
                                                                             Fábio Porto

       WSMO Top Level Elements

                          Objectives that a client may have
                          when consulting a Web Service

Provide the formally                                          Semantic description of
specified terminology                                         Web Services:
of the information                                            - Capability (functional)
used by all other                                             - Interfaces (usage)

                        Connectors between components with
                        mediation facilities for handling
                                                                                               Fábio Porto

Layered Architecture


 Domain     Web Service
                          Mediator   Goal
 Ontology   Description

                   Web Service
                                                           Fábio Porto
WSMO 1.0


   Used as data model throughout WSMO
   Ontology elements: concepts, relations, functions,
    axioms and instances
   Ontology Specification Language: WSML
      –   Web Compatibility:
              Namespaces
              WWW Identification Concept (URI, Literal,
              Basic Datatypes from XML Schema
                                                     Fábio Porto
WSMO 1.0

   De-coupling of Request and Service
      Objective description independent of service usage
      inherent support for discovery & service usage
   Constituting description elements:
      Post-condition: object of interest (computational
      effect: conditions that have to hold after resolution
       (real world aspects)
      => Only objective specification without regard to
       resolution by service
   Usage:
      Goal Ontologies (pre-existing Goal Templates)
      Goal Resolution Process „open‟ to implementations
                                                                                          Fábio Porto
         WSMO 1.0
         Web Services
         - complete item description
         - quality aspects                                - Advertising of Web Service
         - Web Service Management                         - Support for WS functional Discovery

       Non-Functional Properties                                  Capability

            Core + WS-specific                               functional description

                                                                                   Realization of
Interaction Interface                                                              WS by using
for consuming WS                     Web Service                           WS
                                                                                   other Web
- Messages                         Implementation                                  Services
- External Visible                     (not of interest in Web                     - Functional
  Behavior                              Service Description)
- „Grounding‟                                                                      - WS

                    Choreography --- Interfaces --- Orchestration
                                                            Fábio Porto
WSMO 1.0
Web Service Capability
   Non-Functional Properties
   Imported Ontologies
   Used Mediators
      OO Mediator: importing ontologies as terminology
      WG Mediator: link to a Goal that is solved by the
     Web Service
   Pre-Conditions
      Input with conditions that web service expects in
     order to be able to provide its service
     (computational aspects)                                before
   Assumptions
      Conditions that have to hold before the Web
     Service can be executed
      (real world aspects)
                                                               Fábio Porto

Web Service Capability

   Post-Conditions
     Result / Output of Web Service in relation to
    the input, and conditions on it (computational aspects)
   Effects                                                      after
     Conditions / Changes on the state of the world
    that hold after execution of the Web Service (real world
                                                                               Fábio Porto

          WSMO Service Example (1)
  A WSMO Ontology:
#>                                                           A WSMO Service:

  concept route
     dc:description hasValue “Route between two stations”
  startLocation ofType tc:station
  endLocation ofType tc:station

  concept reservation
      dc:description hasValue “Reservation with a
reservation owner”
    reservationNumber ofType xsd:integer
    reservationRoute ofType route
    reservationHolder ofType prs:person
                                              Fábio Porto

     WSMO Service Example (2)
Service Capability:   Service Choreography:
                                                                                    Fábio Porto
    WSMO 1.0

   For handling heterogeneity

       Source                WSMO Mediator
     Component                                                            Target
                          uses a Mediation Service via            1     Component
                 1 .. n

                                         - as a Goal
                                         - directly
                                         - optionally incl. Mediation


   Mediator Types: OO, GG, WG, WW
                                                              Fábio Porto
    WSMO 1.0
    OO Mediator Example
                   Merging two ontologies

Train Connection         OO Mediator
  Ontology (s1)         Mediation Service
                                                     Train Ticket
  Purchase                   Goal:                Purchase Ontology
 Ontology (s2)      “merge s1, s2 and s1.ticket
                    subConceptOf s2.product”


                                                       Fábio Porto
WSMO 1.0
Non-Functional Properties

   Every WSMO element is described by properties that contain
    relevant, non-functional aspects of the item

   Core Properties (for every WSMO element)
     -     Dublin Core Metadata Set
     -     Version

   Service Specific Properties:
     -     Quality of Service Information
     -     Financial / contractual properties
                                                                                           Fábio Porto
  WSMO 1.0
  nfp: Core Properties
ontology <"">
 dc:title hasValue "International Train Connections Ontology"
 dc:creator hasValue "DERI International"
 dc:subject hasValues {"Train", "Itinerary", "Train Connection", "Ticket"}
 dc:description hasValue "International Train Connections"
 dc:publisher hasValue "DERI International"
 dc:contributor hasValues {"Michael Stollberg",
 dc:date hasValue "2004-10-08"
 dc:type hasValue <"">
 dc:format hasValue "text/html"
 dc:identifier hasValue <"">
 dc:source hasValue <"">
 dc:language hasValue "en-US"
 dc:relation hasValues {<"">,
 dc:coverage hasValue "ID:7029392 Name:World"
 dc:rights hasValue <"">
 version hasValue "$Revision: 1.6 $"
WSMO 1.0
                                             Fábio Porto
nfp: Service Specific
   Quality Aspects and other non-functional
    information of Web Services:
       Accuracy              Robustness
       Availability          Scalability
       Financial             Security
       Network-related QoS   Transactional
       Performance           Trust
                 Fábio Porto

WSML: Overview
                                               Fábio Porto

Web Service Modeling Language

   Four elements of WSMO:
    –   Ontologies, Goals, Web Services, Mediators

   WSML provides a formal grounding for the
    conceptual elements of WSMO, based on:
    –   Description Logics
    –   Rule Languages
    –   First Order Logic
                                                   Fábio Porto


   WSML distinguishes between a conceptual
    and a logical expression syntax.
    –   The conceptual syntax is used for the modeling of
        web services, goals, mediators and ontologies.
    –   The logical expression syntax is used for the
        specification of axioms and constraints in an
        ontology and inside the pre- and post-conditions
        of goals and web services.
                                            Fábio Porto


   Based on Logic Programming-variant of F-
    Logic and HiLog
   Minimal model semantics
   Implements default negation
   Allows unrestricted use of function symbols
   Full support for goal/web service modeling
                   Fábio Porto

QoS Base example
Functional Discovery
                                                                   Fábio Porto

       Overall Discovery Process

                  Requester Desire

                                                                         Ease of

  Predefined          Goal              Selected predefined Goal
  formal Goal       Discovery

                  Requester Goal           Goal refinement

Available WS
  Abstract      Web Service Discovery


                                           Service Discovery
                 (possibly dynamic)

                Still relevant WS        Service to be returned
                                                                                              Fábio Porto

                                  Discovery Techniques
                                       Aim of Discovery: detect suitable Web Services
                                        to solve a Goal

                                       different techniques usable
                                          –   Key Word Matching √
                                               match natural language key words in resource
                    Possible Accuracy
Ease of provision

                                          –   Controlled Vocabulary √
                                               ontology-based key word matching
                                          –   Logical Semantic Resource Descriptions √ √ √
                                               … what WSMO aims at
                                                     Fábio Porto

Matchmaking Notions

   Exact Match:
      G, WS, O, M ╞ x. (G(x) <=> WS(x) )

   PlugIn Match:
      G, WS, O, M ╞ x. (G(x) => WS(x) )

   Subsumption Match:
      G, WS, O, M ╞ x. (G(x) <= WS(x) )

   Intersection Match:
      G, WS, O, M ╞ x. (G(x)  WS(x) )          X

   Non Match:
      G, WS, O, M ╞ ¬x. (G(x)  WS(x) )

                                            =G       = WS
QoS-Enabled Semantic WS

          Our contribution on

          In collaboration with
          Le-Hung Vu,
          Manfred Hauswirth, Othman
          Tajmouati, Sebastian Gerlach
                                                                             Fábio Porto


   QoS is critical for users in business applications:
     –   Availability of the service
     –   Execution time
     –   Response time                                           Network/ Software-related
                                                                 QoS parameters
     –   Robustness: error and exception handling
     –   Accuracy of the results (computational services)

… and also a key to business success:
     –   Freshness/Coverage (business information services)
     –   Timely delivery (online book-stores)
                                                                 Application-level QoS
     –   Quality of food (online-order restaurants)
     –   Quality of accommodation (hotel reservation services)

   Quality will enable the ranking of Web service search results in
    the case of many functionally equivalent services
    (compare to Web search!)
                                                                     Fábio Porto

QoS-based WS discovery
   QoS-based WS Matchmaking
    WS providing the same functionality (resulting from functional
            e.g. services providing Stock Market news
    are further filtered out through QoS matchmaking and ordered by
       QoS-based ranking
            e.g. differences in Data freshness
   QoS-based ranking
    –   Service providers may exaggerate their advertised QoS ratings for
        attracting more customers
            e.g. inaccurate data freshness values
    – Users may cheat when rating QoS
     A reputation-based QoS ranking provides some guarantee that a
      customer is not misled by false QoS claims
                                                                  Fábio Porto

    Motivation Example
                                                  data freshness= 3 min
                                                  availability ≥ 0.9999
                           WSGetNews              price=10 euros
Get Stock Headlines :
                                       xignite availability ≥ 0.9999
                                                    data freshness=
                                              data freshness= 3 min 1 min
 data freshness ≤ 5 min
                                            availability ≥ 0.9999
and availability > 0.99
                          WSGetNews                 price=200
                                              price=10 euros euros
and price < 100 euros
                              WSGetNews             data freshness= 1 min
                                      xignite      data fresh ness= 5 min
                                                    availability ≥ 0.9999
                                                   availability ≥euros
                                                    price=200 0.9999
                                                   price=20 euros
                               WSGetQuote       data freshness= 5 min
                                                 data freshness= 5
                                                availability ≥ 0.9999 min
                                                 price=20 euros
                                                price=20 euros

                                                data freshness= 10 min
                                                price=60 euros
                                                      Fábio Porto

     Motivation Example

       Weird, Verireputation
       indicates Xignite/ WSGetNews
       data freshness ≥ 6 min

                              WSGetNews   data freshness= 5 min
                                          availability ≥ 0.9999
                                          price=20 euros
                                 Fábio Porto

Goals (in DIP)

 Model web service QoS
 Specify semantic matchmaking and
  ranking based on the QoS model
 Specify and implement Reputation
  Assessment Mechanisms
 Design a scalable solution
                                                             Fábio Porto

QoS definition
   QoS in DIP is the set of non-functional
    characteristics and environmental conditions
    that distinguishes a service from others
    offering the same functionality

   QoS in Semantic Web Service Discovery
    –   Quality characteristics of concrete services
            How fresh stock market information is?
    –   Quality characteristics of Web services
            Is the web service available 90% of the time?
                                                                            Fábio Porto

 QoS differ individual services

  In theory, a QoS specification should distinguish individual
services, such that a value can be associated to each one of them,
     Giving, s  service, val ( s)  quality  level ( s)  function( s),
     price( s )  val ( s)  profit   , d ( s )  d (q)  d ( f )

  QoS Description

                                                             Functional Description
3. QoS Modelling
                                                       Fábio Porto

    QoS Overall Model

                  QoS upper ontology

                     is the basis for

User QoS Domain        ranking          Service QoS Domain
    Ontology           ontology               Ontology
       used in                             used in
Goal QoS spec.                           WService QoS spec.
  Ontology                                   Ontology
                                                              Fábio Porto

Qos Modeling

   QoS Upper ontology
    –   QoS price model
    –   QoS concepts model
    –   QoS Environmental concepts
    –   QoS comparison model
    –   QoS measurement Model
    –   Relationship between QoS and Environmental concepts
   QoS Domain ontology
    –   Extends basic Upper ontology definitions
            Ex: QoS model for: financial market, tourism,…
   QoS on service description and Goals
    –   QoS instances
    –   Environmental criteria
                                                                                                                                                                   Fábio Porto

  QoS Upper Ontology
                                                                                              1                 1
           PriceModel                                          MeasurementUnit                                        UnitChangingModel
        +hasUpperBound                                                                        1                 1    +value
                                                                                                      -source unit
        +hasLowerBound                                                                                               -refUnit
                               1                      1
                                                                                                        -base unit
                                                                                                                     MeasurementModel                      MeasurementType

                                                                                                             1..*                         1            1

                                                                             1        1
-providedUnder     1                                                                                                 QoSComparisonModel                    ComparisonResult
                               -composes of                                                       1             1
                                                                                                       -1st param
        QoSSpecification                                                QoSParameter
                                                                                                                                               1       1
                                                                      +hasLowerBound              1   -2nd param
                               1                               *      +hasUpperBound

                                                                             1   1    1                         1
        ContextualDependency                                                                                          QoSMatchingModel                     MatchingResult
                                       -hasContextualFactors                                                   1
                                       1                                                              -1st param
                                                                                                                                              1        1
                                       1                                                              -2nd param
                                                                                          1                     1
                                                                                                       -1st param
                                   -composes of                    ContextualFactor
        ContextSpecification                                                                                                                       1
                                                               +hasLowerBound             1           -2nd param
                                   1                      *
QoS Upper ontology                                        Fábio Porto

   Price model
    –   Specify price unit
    –   Price rules
    –   Conversion functions
    –   Relationship between the price concept and other QoS
   Comparison Model
    –   Comparison criteria between QoS Concepts
            Higher better, lower better
            Inclusion
            Similarity,…
   Measurement model
    –   Measurement units
    –   Conversion rules
                                Fábio Porto

Snippet of QoS Upper Ontology
Web Service and Goal                                                        Fábio Porto

QoS Model
   Composed of two parts:
    –   QoS criteria
            Express the space of QoS values accepted/required by the
             Web service or goal, respectively (Quality level)
               –   A constraint over QoS concept instances
            A formula of the type: (X1ν C‟1) ^ (X2ν C‟2) ^ … ^ (Xnν C‟n)
               –   Where C‟I is a constraint over instances of CI, a QoS concept,
                   and Xi is either ( Cif C i' is optional, or false otherwise)

            Example:
               –   Throughput > 1MB/s and fileSize <= 100MB
    –   Environmental criteria
            Specifies conditions the user must agree in other to be
             eligible for a certain quality level
            It is expressed similarly to a QoS criteria formula
            Example:
               –   Location=„Lausanne‟ and link=„wireless‟
                                                           Fábio Porto

QoS specification in WSML

   Defined per interface
   Specified as a separate ontology imported by the
    web service or goal interface
    –   Work around to fit within current WSMO NFP limitations;
   QoS ontology in WS and goal can import domain
    specific QoS ontologies
                                               Fábio Porto


              Goal/WS description

     Interface desc.

       0..1                         1..1
        QoS Spec         Required capability

       0..n                         0..n
          QoS             Environmental
       Value spec          specification
Snippet of a WS QoS Spec.   Fábio Porto

                                                     Fábio Porto

Ranking ontology

   Specifies:
    –   Ranking model
            Associate values to matching result
            Matching thresholds
    –   User preferences
            Users weight differently QoS Concepts
    –   Reputation information
            User relevance for reputation info
                                Fábio Porto

Code snippet ranking ontology
4. QoS-based matchmaking
    QoS Overall Model                                       Fábio Porto

    with matching

                    QoS upper ontology

                       is the basis for

    Domain               ranking               Domain
  QoS ontology           ontology            QoS ontology
       used in                               used in
SWS description                              Goal description
  ontology                                      ontology
                  compose Knowledge Base

                                           QoS service
                          ranking          discovery
                                                       Fábio Porto

  Matchmaking Model

For matching:
         KB  Uo           Do        Ro

Matching is:    KB  KB        WSqo         Gqo
       For QoS concepts:
      k , KB   (Cpk , Cuk ,Wqos),Wqos  Mqos
     For environment:
      k , KB  env (Cuk , Cpk ,Wenv ),Wenv  M env
                                                                              Fábio Porto

         QoS-based matchmaking
          Goal description                              WS description

Interface desc.                                                      Interface desc.

  0..1                        1..1               1..1                            0..1
QoS requirement     Required capability       Provided capability        QoS offering

  0..n                        0..n                 0..n                          0..n
     QoS             Environmental              environmental                QoS
 required spec        specification              requirement             provided spec

                                contextual matchmaking

                             QoS parameter matchmaking
                                                                    Fábio Porto

QoS-based Service Ranking
   The QoS-based ranking of a service depends on
     –   Compliance (i.e. match) of service‟s QoS definition with user‟s
         QoS goals (QoS ontology)
     –   Reputation-based estimates for service‟s QoS attributes
     –   User preferences QoS attribute aggregation (Ranking ontology)

   The reputation-based QoS estimation depends on
     –   QoS ratings reported by users
     –   QoS ratings reported by Trusted Agents (e.g., AlertSite, Dot-
         Com Monitor, Empirix)
     –   QoS ratings advertised by the service providers

   This work is among the first ones applying reputation-based
    mechanisms for QoS-based service discovery [VHA05]

   The reputation-based QoS ranking is compliant with the more
    general trust model developed in D4.21.
                                                       Fábio Porto

Dealing with partial matching

   When comparing two QoS instances (QoS or
    environment concept) the comparison model
    associates a value within a matching range;
   We define in the user QoS ontology
    matching thresholds:
    –   Upper bound: εh
    –   lower bound: εl ≤ min( (1- εh ), εh )
    –   For 0 ≤ ε ≤ 1
                    εl           εh                1

        Not match
                         partial match   exact match
                                                     Fábio Porto

                     UploadSpeed Download     Supporte
                     (q1)        speed(q2)    d size (q3)

 Goal                >25 KB/s      >25KB/s    ≤ 500 MB

 S1                  5-20 KB/s     30-40 KB/s ≤ 500 MB

 S2                  20-30 KB/s    40-80 KB/s ≤ 1.5 GB

        Weight: w1=0, w2=1, w3=1
        εl= εh=0.5
  5. QoS-based discovery
framework and architecture
QoS-based                                          Fábio Porto

service discovery framework

   The whole QoS-based discovery framework is
    modeled in term of algebraic operators which
    enables plugging-in of different implementations

 Highly customizable implementation
                                                                                                        Fábio Porto
       QoS Discovery Architecture
                                            Service                Service
                                             User                  Provider

• QoS discovery architecture
                                          DIP Discovery Component
   • QoS discovery component
       • QoS matchmaking

                                                                                                                 D i s c o v e r y S e r v i c e In t e rf a c e
                                                                                                                     S e m a n t ic W e b S e r v ic e
       • Ranking based on QoS
       • Functionality implemented                          WSMO Service

       as algebra operators
                                                                                                           Functional -based
       DIP discovery API                                                                                      Discovery
                                                                                                                 ( D 4 .8 )

   • Derby DBMS stores
                                     Mediation Service Interface              WSMO Registry Interface

       • WSMO collections
                                                        DIP -WSMX components

       • QoS estimates
                                                                                                        Fábio Porto
      QoS Discovery Architecture
                                            Service                Service           QoS
                                             User                  Provider         Reporter

• Reputation
   – Computes reputation estimates        DIP Discovery Component

   – Apply discovery over computed
                                                              QoS                   QoS Data

                                                                                                                 D is c o v e ry S e rvic e Int e rf ac e
                                                                                                                    S e m a n t ic W e b S e r v ic e
                                                            Reputation              Repository

                                                            WSMO Service

                                                            Discovery                                      Functional -based
                                                                                                                 ( D 4 .8 )

                                     Mediation Service Interface              WSMO Registry Interface

                                                        DIP -WSMX components
                                                                                                       Fábio Porto
      QoS Discovery Architecture
                                           Service                Service           QoS
                                            User                  Provider         Reporter

• Parallel Discovery
   – Scales-up with parallel             DIP Discovery Component
                                                             QoS                      QoS Data

                                                                                                                D is c o v e r y S e r v ic e In t e rf a c e

                                                                                                                   S e m a n t i c W e b S e r v ic e
                                                           Reputation                 Repository
   – Efficient adaptive parallel   Optimizer

                                    manager                  Internal
                                                           WSMO Service

                                     QoS                   QoS                    QoS
                                   Discovery             Discovery              Discovery                 Functional -based
                                                                                                                ( D 4 .8 )

                                    Mediation Service Interface              WSMO Registry Interface

                                                       DIP -WSMX components
                              Query Execution Plan for                                                                                                                                                Fábio Porto

                              Processing Service Discovery

                 DBMS         Scan


                                                                        services possibly matches
                                                                        QoS requirements
 User query
                                             user query with                                                                                       service
                                             Bloom key                                                                                                                                                         Ranked list of
                                                                                                            service               QoS                           service              service                   matched services
                            BloomKey                                                                                                                    Merge             Rank                       Project
                                                                                             HashJoin                 Split

                        concept group                                                                                              ...
                                                                                                                                                                             User preferences
DBMS      Scan
                                                                                  service matching
                                                                                  functional requirements
                                                                                                                                     User query,
                                                                                                                                     user preferences
                                        service         Functionality
              DBMS         Scan
                                                          match                                                                                                                           Data flow for input/output

                                                               User query,                                                                                                                Data flow for configuration
                                                               user preferences

                                                                                                                                                                                          Operator that could be parallelized

                                                                                                                                                                                 A       Content of the data flow
                                                          Fábio Porto

   A new model for QoS in SWS
   First work to introduce QoS in DIP/WSMO!!
   A matchmaking and ranking approach
   Is completely configured on a domain and
    user/provider basis
    –   Including configuration of
            Domain ontology
            Ranking ontology
            Preferences
   It is completely implemented
   More information:
    –   QoS-enabled Semantic Web Service Discovery: a
        Personalized approach, submitted to IDEAS 2007;
Reasoning on Dynamically Built
Reasoning Space with Ontology

      Fabio Porto
      École Politechnique Fédéral de Lausanne
      Database Laboratory
                                 Fábio Porto


   Introduction
   Preliminaries
   General Framework
   Reasoning space
   Reasoning algorithm
   Conclusion and Future work
                                                       Fábio Porto


   Ontologies are increasingly used as explicit
    models of the conceptualization of underlying
    information sources;
   Different ontologies:
    –    representing partially intersecting domains
    –   same domain observed from different
   Applications require reasoning over such
    autonomously developed ontologies
                                                                           Fábio Porto

Multiple ontology scenarios

   In e-science:
    –   In order to study colon carcinoma disease, a biologist would
        conjointly use its own ontology with others such as:
            medical ontology (UMLS), anatomical ontology(mouse
             ontology, HUMAT), Pharmacogenomics (PharmGKB) and GO;
   In e-business:
    –   In automatic web service discovery within a virtual travel
            Location ontology, currency ontology, flight ontology, date
                             Fábio Porto

Overall picture



                                                                     Fábio Porto


   User group agrees on a common ontology but trusts on
    information defined on other ontologies;
   Autonomously developed ontologies:
    –   Partially intersect;
    –   Represented through different logical languages;
    –   Require mappings among ontologies models;
    –   May include contradictions;
   Current reasoners:
    –   Consider ontologies forming a single logical model;
    –   Need different ontologies to be aligned;
   Centralized Reasoning:
    –   Transferring large ontologies to a central site is costly;
                                                     Fábio Porto

Problem Statement

   Given a set of autonomously developed
    ontologies linked by mapping
    correspondences to a group ontology,
    conceive a reasoning strategy compatible
    with current centralized reasoners
   Contribution
    –   Strategy to dynamically build a reasoning space
        based on: a query, an ontology space and
                                                                        Fábio Porto

   Ontologies expressed in SHIQ:
    –   Distinct sets of concepts, roles and individuals;
    –   ,  are concepts and, if C,B are concepts, then C, ( C  B), (C
         B) are concepts;
    –   An ontology O is modeled by an interpretation I.
          I=(I, (·I))
          Concept names are interpreted as subsets of I
          Complex expressions are interpreted according to the
            following equations:
               –  I= I;  I= ; ( C  B)I= (C I BI); ( C  B)I= (C   I
               – C I= I \ C I
            Knowledge Base:
               –   ( C  B), ( R  S), a:C, <a,b>:R
            Interpretation
               –   ( C  B) iff ( C I  B I )
               –   a:C iff aI  C I
                                                                                             Fábio Porto

General Framework

   Ontology Space OS={O1, O2,…, On}, where Oi is an ontology;
   Ontology Module Mid=<id,D,L,Ob,Cid,OS>
     –   C correspondences(bridge rules in c-owl):
              Oi:C ≡ Oj:D; Oi:C  Oj:D ;Oi:C  Oj:D ; Oi:R ≡ Oj:S
              Oi:v ≡ Oj:t;
                 –   Where C, D are concepts; R, S are roles and v,t are instances   Peer Pi=<Mi,S>
   Query:
   Q= q1  q2  …  qt. [Horrocks,Tessaris 2000]
     –   qi,1≤ i ≤ t, is a term : x:C or <x,y>:R
     –   qi is satisfied by O iff O qi
   Boolean queries:
     –   CD
   Peer: P=<M,Reasoner,query rewriter>
                                                                             Fábio Porto

      Reasoning Space

      Virtual ontology built to answer a query Q
       with relevant entities from OS
        –   RS  {OS  C}
      How to compute relevant entities (Q,OS)?
        –   Relevant entity e in Oi
                 ei  Oi is relevant to Q iff  ej in qj ,term of, Q, such that
                    –   ei  ej 
RS                                                   C1    C2 e
        C2                                                     i
 C1                             ei ej                   CN  R1
      CN R1
                           qi           Oj
                                                                             Fábio Porto

      Reasoning Space

      Virtual ontology built to answer a query Q
       with relevant entities from OS
        –   RS  {OS  C}
      How to compute relevant entities (Q,OS)?
        –   Relevant entity e in Oi
                 ei  Oi is relevant to Q iff  ej in qj ,term of, Q, such that
                    –   ei  ej 
RS                                                          C2      RS„
        C2                      ei   ej
 C1                                                       CN R1
      CN R1
                           qi             Oj
                                                     Fábio Porto

Building Reasoning Space

   Successively extend the Reasoning space by
    identifying relevant ontology entities:
    –   Initially assumes RS=local ontology
    –   Apply a reasoning space extension function that
        for each ontology in OS, identifies relevant
   Reason over the current RS
                                                                           Fábio Porto

           Getting relevant entities
          x:Protein                                      Bodily Process
(x,lactation):involved_with 

                                                          Lactation   digestion

                                                         Disease process

                  Receptor Protein
                                                          Breast Cancer
                                         Involved_with                Stomach Cancer
                                                                           Fábio Porto

           Getting relevant entities
          x:Protein                                      Bodily Process
(x,lactation):involved_with 

                                                          Lactation   digestion

                                                         Disease process

                  Receptor Protein
                                                          Breast Cancer
                                         Involved_with                Stomach Cancer
                                                                               Fábio Porto

           Extending RS
          x:Protein 
(x,lactation):involved_with 
  (x,disease):involved_with                          Bodily Process

                                     Involved_with       Lactation
                                                             Disease process

                  Receptor Protein
                                                               Breast Cancer
                                                                                   Fábio Porto

             General Framework
Are there proteins involved in lactation and disease processes?

                                P1                                                        O2
          x:Protein 
(x,lactation):involved_with 

                                            O1                                            C21

  (x,disease):involved_with                                        E2

                                            C13        Rewriter

                                                                  q13               C31


                                                                   Fábio Porto

Reasoning space algorithm

   reasonspace(query Q,ontology Ob,OS,correspondence C) : answer
   {     RS´:= {Ob};RS=;
        q:= i=1,t qt ; /* qt terms of query Q */
        q:=q – {satisfied(q)};
       While (q  and RS  RS’) {
          RS´= f(q,RS,OS);
          answer:= answer  {evaluate(q ,RS´)};
          q:=q – {satisfied(q)}; }
       return answer; }
                                                                                Fábio Porto

    Related work

   L. Serafini, A: Tamilin, ´Distributed reasoning services for multiple ontologies´.
    Technical Report DIT-04-029, University of Trento, 2004
   M. Lenzerini,”Data Integration: A Theoretical Perspective”, ACM PODS 2002;
   D. Calvanese, G. De Giacomo, M. Lenzerini, and R. Rosati, Logical foundations
    of peer-to-peer data integration, PODS 2004;
   E. Fancioni, G. Kuper, A. Lopatenko, L. Serafini, A Robust Logical and
    Computational Characterisation of peer-to-peer Database systems, DBISP2P-
    2003, co-loacated VLDB2003;
   K. Alberer, P. Cudré-Mauroux, M. Hauswirth,” GridVine: Building Internet-Scale
    Semantic Overlay Networks
    ”, The 3rd International Semantic Web Conference (ISWC2004), Hiroshima, 7-
    11 Nov 04
   J.Lin and A. O Mendelzon. Merging databases under constraints, Intl. J. of
    Cooperative Information Systems, 7(1):55-76,1988.
   A. Halevy et al., Z.G. Ives, D. Suciu, I.Tatarinov. Schema mediation in peer
    data management systems. ICDE 2003.
   L. Serafini, F. Giunchiglia, J. Mylopoulous, P. Bernstein. Local relational model:
    A logical formalization of database coordination. In context 2003, 2003.
                                                                     Fábio Porto

Conclusion and Future work

   Preliminary results on reasoning over autonomously developed
    distributed ontologies;
   Present a strategy that:
    –   Uses current reasoner technology;
    –   Reduces the cost associated to transferring ontologies;
    –   Identifies contradictions among ontologies;
    –   Is based on database approach for evaluating distributed queries;
   Future work:
    –   Use query results to update mapping information;
    –   Evaluate the approach comparing to distributed reasoning
        strategies based on distributed tableau method;
    –   Evaluate quality of results;
    –   Conceive an approach for determining which peer to ask
Thank You !!!
4. Reputation-based QoS ranking
                                                                                 Fábio Porto
          QoS Specification in WSMO Web
webService FileHostingService
    capability ….
    interface freeService
        nfp dc#relation hasValue freeService#serviceSLA endnfp
     interface subscribedService
        nfp dc#relation hasValue subQoS#serviceSLA endnfp
axiom freeService#serviceSLA definedBy
  serviceSLA(any,qos#yes) impliedBy ?userExecEnvironment[qos#hasNetworkConnectionSpeed
   hasValue ?networkUser,            qos#hasLocation hasValue ?locationUser] memberOf
   and qos#envRequirementMatch(?networkUser,requiredNetworkConn,qos#exactMatch) and
   qos#envRequirementMatch(?locationUser,qos#Switzerland,qos#exactMatch) .
instance serviceSLA --
instance providedUploadSpeedSLA …
instance providedNumberConcurrentUploadsSLA …
Specification of QoS Requirements in                                     Fábio Porto

WSMO Goal Description
Goal RequiredFileHostingService
   capability …
      nfp dc#relation hasValue goalQoS#satisfiesQoS endnfp
axiom goalQoS#satisfiesQoS definedBy
   satisfiesQoS (fileHostingQoS#UploadSpeed, qos#yes) impliedBy
     ?serviceQoSSpec[fileHostingQoS#hasUploadSpeed hasValue ?serviceUploadSpeed]
     fileHostingQoS#FileHostingServiceQoSSpecification and
     qos#qosRequirementMatch(?serviceUploadSpeed,reqUploadSpeed,qos#exactMatch) .
   satisfiesQoS (fileHostingQoS#NumConDownloads, qos#acceptable)
    impliedBy …
   satisfiesQoS (fileHostingQoS#SupportFileSize, qos#yes) impliedBy …
instance reqUploadSpeed memberOf fileHostingQoS#UploadSpeed …
instance reqNumConUploads memberOf
    fileHostingQoS#NumberConcurrentUploads …
instance reqSupportFileSize memberOf fileHostingQoS#SupportFileSize …
                                                            Fábio Porto

     QoS Modelling
                    QoS upper ontology

                       is the basis for

  Web service            ranking                Goal
  QoS ontology           ontology            QoS ontology
       used in                               used in

SWS description                              Goal description

                  compose Knowledge Base

                                           QoS service
                          ranking          discovery
                                                            Fábio Porto

     QoS Modelling
                    QoS upper ontology

                       is the basis for

  Web service            ranking                Goal
  QoS ontology           ontology            QoS ontology
       used in                               used in

SWS description                              Goal description

                  compose Knowledge Base

                                           QoS service
                          ranking          discovery

Shared By: