MAFRA – A MApping FRAmework for Distributed

OntologiesAlexander Maedche, Boris Motik, Nuno Silva and

Raphael VolzFZI, University of Karlsruhe

Presented byFrancisco Martin-Recuerda

Digital Enterprise Research Institute (DERI)

2004-01-20 WP4 - Ontology Mediation 2

Outline

• MAFRA• Conceptual Framework• Semantic Bridging• Example

2004-01-20 WP4 - Ontology Mediation 3

MAFRA

Framework for mapping distributed ontologies implemented within KAON (Ontology Management tool developed by University of Karlsruhe):

– Horizontal dimension: defines the phases in a mapping process (Lift &Normalization, Similarity, Semantic Bridging, Execution and Postprocesing).

– Vertical dimension: defines the modules that interact with the elements of the horizontal dimension (Evolution, Domain Knowledge & Constraints, Cooperative Consensus building and GUI).

2004-01-20 WP4 - Ontology Mediation 4

MAFRA

2004-01-20 WP4 - Ontology Mediation 5

Conceptual Framework – Horizontal Dimension

• Lift & NormalizationDefine a uniform representation (in RDF(S)) in order to normalized the ontologies that we want to map. In the proccess syntax differences are eliminated and the semantic diferences are slighly reduced. The result is a list of normalized lexica.

• SimilarityMulti-strategy proccess that calculated similarities between ontology entities using different algorithms.

• Semantic BridgingRelate semantically entities from the source and target ontologies encapsulating all necessary information to transform instances of one source ontology entity to instances of one target ontoogy entity .

2004-01-20 WP4 - Ontology Mediation 6

Conceptual Framework – Horizontal Dimension (II)

• ExecutionThis module actually transforms instances from the source ontology into target ontology by evaluating the semantic bridges defined earlier

• Post-processingTake the results of the execution module to check and improve the quality of the transformation results (like object identity: recognize that two instances represent the same real-world object).

2004-01-20 WP4 - Ontology Mediation 7

Horizontal Dimension - Similarity

1.Lexical similarityUsing Wordnet and altered Resnik algorithm, the strategy focuses on map terms with lexical similarities.

2.Property similarityMap concepts with similar properties (attributes and/or relations)

3.Bottom-up similarityThe algorithm tries to discover new similarities analizing mapping elements of the taxonomy from the lower level to upper level .

4.Top-down similarityIt is like the previous one, but it starts from the upper level to the lower level

2004-01-20 WP4 - Ontology Mediation 8

Horizontal Dimension – Semantic Bridging (Steps)

1. Stablish a mapping of entities to be bridge.(based in the similarities founded in the previous phase).

2. Specify matching properties for each concept bridge.

3. Look for no target concepts. (It is not possible to find a target for a source concept, the algorithm looks for if there are source superconcept mappings and map the source concept in the same way).

4. Improve quality of bridges between source sub-/concepts and target concepts.

5. Associate transformation procedures to translations.

(define how translates source instances into target instances) .

2004-01-20 WP4 - Ontology Mediation 9

Horizontal Dimension – Semantic Bridging (Nature)

• Entity dimensionSBs may relate ontology entities (concepts, relations, attributes and extensional patterns (modeling the content of the instance)).

• Cardinality dimensionDetermine the number of ontology entities at both sides of the semantic bridge.

• Structural dimensionDefine how semantic bridges may be combined into more complex bridges.

• Constraint dimensionDefine conditions that must hold the execution module when transforms instances from the source ontology into the target ontology.

• Transformation dimensionReflect how instances of the source ontology are transformed during the mapping process.

2004-01-20 WP4 - Ontology Mediation 10

Horizontal Dimension – Semantic Bridging Ontology (SBO)

• Semantic Bridging Ontology (SBO)Specification (using DAML+OIL) of all available semantic

bridges organized in a taxonomy.

– Classes Concepts, Relations and AttributesRepresent the entities of the mapping ontologies.

– Class Semantic BridgeDefine the relations to source and target entities.

– Class ServiceReference resources that are responsible to connect to or describe transformations.

– Class RuleRepresent constraint and transformation-relevant information.

2004-01-20 WP4 - Ontology Mediation 11

Horizontal Dimension – Semantic Bridging Ontology (SBO) (II)

– Class TransformationAssociate with a transformation procedure a set of extra requirements.

– Class ConditionRepresent the conditions that should be verified in order to execute a semantic bridge

– Class CompositionAllow to any semantic bridge to aggregate many different bridges.

– Class Alternative Group several mutual exclusive semantic bridges.

2004-01-20 WP4 - Ontology Mediation 12

Horizontal Dimension – Semantic Bridging Ontology (SBO) (III)

2004-01-20 WP4 - Ontology Mediation 13

Conceptual Framework – Vertical Dimension

• EvolutionSyncrony the changes in the source and target ontologies with the semantic bridges defined by the Semantic Bridge module.

• Cooperative Consensus BuildingFrom multiple alternative possible mappings the tool help to stablish a consensus between several proposals of people involved in the mapping task.

• Domain constraints and Background Knowledge. The tool allow users to include estra-information (glossaries to help indentify synoms, lexical ontologies like Wordnet) to improve the improve the quality of the mapping.

• GUIMake the work to visualize the elements of the source and target ontologies much easier in the same way that the semantic bridges stablished to represent the mapping between entities.

2004-01-20 WP4 - Ontology Mediation 14

Example

2004-01-20 WP4 - Ontology Mediation 15

Example