A Classification of Ontology Change by wuyunyi

VIEWS: 28 PAGES: 1

									                                                    A Classification of Ontology Change
                                                                    1,2                                                                   2                                                             2
                          Giorgos Flouris , Dimitris Plexousakis , Grigoris Antoniou
                                                                      georgios.flouris@isti.cnr.it, {dp, antoniou}@ics.forth.gr
                                           1 CNR,   Istituto della Scienza e delle                         2 FORTH,     Institute of Computer Science
http://www.isti.cnr.it/                         Tecnologie dell’ Informazione                                     Information Systems Laboratory                                                  http://www.ics.forth.gr/isl
     Ontology Change Definition                                                        Motivation for this Paper                                                           Ontology Change Subfields
        Ontology change is the problem of                   Situation: Due to its complexity, the problem is being addressed by several different,                    1. Ontology    Evolution
        deciding the modifications to perform               but closely related, interlinked and often overlapping research disciplines; each of these                2. Ontology    Versioning
        upon an ontology in response to a                   disciplines deals with a certain facet of the problem from a different view or perspective,               3. Ontology    Mapping
        certain need for change, as well as                 covering different application needs, change scenarios or needs for change                                4. Ontology    Morphism
                                                                                                                                                                                                                        Heterogeneity
        the implementation of these                                                                                                                                   5. Ontology    Alignment
                                                            Problems:      Overlapping disciplines; boundaries not clear; confusing use of terms                                                                         Resolution
        modifications and the management                                                                                                                              6. Ontology    Articulation
        of their effects in depending data,                 Purpose of this paper:                                                                                    7. Ontology    Translation
        ontologies, services, applications,                            Make broad survey and study relationships/boundaries between subfields                        8. Ontology    Integration
        agents or other elements                                       Classify ontology change approaches                                                           9. Ontology    Merging
                                                                       Propose a terminology; clarify terms; study alternative literature definitions

                                                                                                            Red
                                  Ontology Evolution at a Glance                                                                                                             Comments on Ontology Evolution




                                                                                                                                Ontology Evolution
        Ontology                  Purpose:       Apply required modifications to an ontology;
                                                                                                             …
                                                                                                                                                                             Implements change(s) to the source ontology
                                                                                                                                                             Red
        Evolution                 modifications are initiated due to changes in the domain of the                                                                            Does not provide ontology versioning features




                                                                                                                                    Algorithm
                                  ontology or its conceptualization
                                  Input:         An ontology and a (set of) change operation(s)            … ………                                                             Initiated to handle changes in the domain (i.e.,
                                                                                                                                                                             when the modeled world has changed) or to
                                  Output:        An ontology                                              Change                                               …             handle changes in the conceptualization (i.e.,
           Definition                                                                                                                                                        when the domain should be modeled differently)
The process of modifying an
                                                                                                      Change Description:                                  … ………
                                  In the figure, a set of Green changes are applied to the Red                                                                               Usually performed in six phases
ontology in response to a                                                                             Change #1
                                  ontology: the Red ontology is changed accordingly in order to       Change #2
certain change in the domain or   accommodate the changes                                             …
its conceptualization


                                                                                                         Blue [Dark]        Blue [Normal]              Blue [Light]


                                  Ontology Versioning at a Glance                                                                                                             Comments on Ontology Versioning
                                                                                                              …                     …                        …
                                  Purpose:     Handle multiple versions of an ontology; provide                                                                               Version ids are required to identify versions
        Ontology                                                                                            … ………              … ………                     … ………
                                  transparent access to the “correct” version, depending on the                                                                               Must determine which changes result to a new
        Versioning                accessing element                                                                                                                           version (and need a new id)
                                  Input:       Different versions of an ontology
                                  Output:      A versioning system                                                                                                            Needs to determine compatibility between
                                                                                                                      Ontology Versioning                                     versions
                                                                                                                           Algorithm                                          Provides transparent access to versions
                                  In the figure, three versions of the Blue ontology (Blue [Dark],
                                  Blue [Normal] and Blue [Light]) are managed by a versioning                                                                                 Useful feature: “version relations”
            Definition
The ability to handle an          system; each of the three accessing applications, needs                                                                                     Useful feature: “deltas” between ontologies
evolving ontology by creating     access to (is compatible with) a different version of the Blue         Light Blue          Dark Blue                 Normal Blue
and managing different versions   ontology; it is the job of the versioning system to provide            Application         Application               Application
of it                             proper (and transparent) access




                                  Heterogeneity Resolution at a Glance
                                                                                                                  Red                                   Blue
                                                                                                                                                                           Comments on Heterogeneity Resolution
                                                                                                                              Translation                                  Change is not explicit but is performed at run-time
                                  Purpose:       Resolve heterogeneity problems between
     Heterogeneity                ontologies; allow interoperability                                              …                                      …
                                                                                                                                                                           (during interoperability and communication)
                                                                                                                                Rules
      Resolution                  Input:         Two (heterogeneous) ontologies                                                                                            Articulation is equivalent to alignment, because a
                                  Output:        Some kind of translation rules between the                       … ………                                … ………               relation can be decomposed into a pair of
                                  source ontologies; these translation rules identify related                                                                              functions from some intermediate source
            Definition            signature elements and/or axioms (depending on the algorithm
                                  type)                                                               The various subfields related to heterogeneity resolution            Ontology Translation does not produce translation
The process of relating two
                                                                                                      are identified based on the type of translation rules                rules; there are two distinct understandings of the
ontologies via some kind of
                                                                                                      produced at the output. More specifically:                           term in the literature:
translation rules in order to
                                  In the figure, a set of translation rules are produced to relate    Mapping:       Function between signatures                           1. Change the representation language of the
mitigate heterogeneity problems
                                  the Blue ontology and the Red ontology; the type of translation     Morphism: Function between signatures and axioms                     ontology
in terms of language, syntax,
                                  rules produced determines the type of the algorithm (mapping,       Alignment: Relation between signatures                               2. Implement a signature mapping upon the
terminology, modelling
                                  morphism, alignment or articulation)                                Articulation:An intermediate ontology, plus functions                ontology (produced by an ontology mapping
conventions etc
                                                                                                      between this intermediate ontology and each source                   algorithm)


                                   Ontology Integration at a Glance                                        Blue                                                            Comments on Ontology Integration

                                   Purpose:     Fuse the information found in two ontologies                                                                               Very similar to ontology merging; the only
                                                                                                                                Ontology Integration




                                                                                                                                                                           difference is in the domain of the sources
        Ontology                   covering loosely related (i.e., similar) domains
                                                                                                             …                                           BlueRed
                                   Input:       Two ontologies (covering similar domains)                                                                                  Mainly applied when the focus is to fuse
       Integration
                                                                                                                                     Algorithm




                                   Output:      An ontology                                                                                                                knowledge from different sources in order to cover
                                                                                                           … ………                                                           a broader domain
                                                                                                            Red                                                …           Very useful in ontology development (integrating
                                   In the figure, the Blue ontology is integrated with the Red
             Definition                                                                                                                                                    independently developed subontologies makes
                                   ontology; the ontologies cover similar domains, so the result
 The process of composing the                                                                                                                              … ………           ontology design more efficient)
                                   (BlueRed ontology) contains each of the sources in loosely
 information found in two or
 more ontologies covering
                                   related (and easily identifiable) modules                                 …                                                             Heterogeneity resolution is a major part of the task
 loosely related domains                                                                                                                                                   of integration (but ontology integration is more
                                                                                                           … ………                                                           than that)


                                   Ontology Merging at a Glance                                            Blue                                                            Comments on Ontology Merging

                                   Purpose:     Fuse the information found in two ontologies                                                                               Very similar to ontology integration; the only
        Ontology                                                                                                                                                           difference is in the domain of the sources
                                                                                                                                Ontology Merging




                                   covering highly overlapping or identical domains
                                   Input:       Two ontologies (covering identical domains)
                                                                                                             …                                            Purple
        Merging                                                                                                                                                            Mainly applied when the focus is to fuse
                                                                                                                                    Algorithm




                                   Output:      An ontology                                                                                                                knowledge from different sources in order to
                                                                                                          … ………                                                            describe the domain more accurately
                                                                                                           Red                                                 …           Very useful when we have a number of ontologies
           Definition              In the figure, the Blue ontology is integrated with the Red
The process of composing the       ontology; the ontologies cover identical domains, so the                                                                                each partially describing the same domain
information found in two or                                                                                                                                … ………
                                   information coming from the source ontologies is greatly                                                                                Heterogeneity resolution is a major part of the task
more ontologies covering highly    intermingled (non-identifiable) in the result (Purple ontology)           …                                                             of merging (but ontology merging is more than
overlapping or identical                                                                                                                                                   that)
domains                                                                                                   … ………
                                                                                                                               This work was carried out during the first author's tenure of an ERCIM “Alain Bensoussan” Fellowship Programme

								
To top