Leipzig 02

Ontologie-Fusion mit Formaler Begriffsanalyse : Ontologie-Fusion mit Formaler Begriffsanalyse Gerd Stumme Institute for Applied Informatics and Formal Description Methods (AIFB) Alexander Mädche Research Center on Information Technologies at the University of Karlsruhe (FZI)


Slide2 : 1 Ontologies and the Semantic Web 2 FCA-Merge 3 Outlook Agenda


Ontologies : Ontologies Ontologies have been widely and successfully applied in the area of information integration, natural language understanding, and are now become a key ingredient of the Semantic Web An example ontology: Project DepartmentManager Person subConcept FORALL X, Y Y: Person[worksIn -andgt;andgt; X] andlt;- X:Project[hasMember -andgt;andgt; Y]. TopConcept cooperates worksIn name ID-IST


Ontology Environment OntoMat : Ontology Environment OntoMat


Ontologies and the Semantic Web : Ontologies and the Semantic Web The current WWW is a great success with respect to the amount of available information the number of users Reasons for this success are among others linked information sources decentralization. However, one problem of the current WWW is that the information may only be interpreted by humans. The Semantic Web tries to overcome this problem by using machine-processable and -interpretable metadata and ontologies.


Our approach : Our approach Ontologies are suitable means to establish „autonomous' and „local' semantics. To establish the Semantic Web vision we require an architecture for federated ontology-based systems supporting the paradigm of decentralization. We adopt ideas from the area of multi-database systems (cf. [Sheth, Larsen 1990]). Important issue for federating ontologies: Merging two input ontologies as a combination of two autonomous systems.  FCA-Merge, a new bottom-up approach!


Federated DB Systems : Federated DB Systems


Federated Ontology-basedSystem : Federated Ontology-based System


Slide9 : 1 Ontologies and the Semantic Web 2 FCA-Merge 3 Outlook Agenda


FCA-Merge: Bottom-Up Merging of ontologies : FCA-Merge: Bottom-Up Merging of ontologies The idea of FCA-Merge is to create, based on the source ontologies, a concept hierarchy - the concept lattice -containing the original concepts. Ontology concepts having the same extension are identified in the concept lattice. Our approach is extensional, i.e., it is based on objects which appear in both ontologies. Concepts having the same extent are supposed to be merged. The knowledge engineer can then create the target ontology interactively, based on the insights gained from the concept lattice.


Generating extensions if necessary : Generating extensions if necessary If we cannot annotate existing objects for that purpose, we will use documents as artificial objects. I.e., concepts which always appear in the same documents are supposed to be merged. As said above, concepts having the same extension are supposed to be merged. If we have objects annotated by both ontologies, we can directly compute the concept lattice. If there are no objects which are annotated simultaneously in both ontologies (e.g., if all Daimler cars and all Chrysler cars are described by both the Daimler and the Chrysler ontology), we have to create such objects first.


FCA-Merge – Method : FCA-Merge – Method O1


The Framework : The Framework uses Propose new concepts/ relations dictionaries/natural language texts


Slide14 : 1 Ontologies 2 FCA-Merge i) Linguistic Analysis and Context Generation ii) Generating the Pruned Concept Lattice iii) Generating the new Ontology from the Concept Lattice 3 Outlook Agenda


Framework : Framework uses Propose new concepts/ relations dictionaries/natural language texts


Information Extraction System SMES (DFKI Saarbrücken) : Information Extraction System SMES (DFKI Saarbrücken) Linguistic Knowledge Pool Lexical database: 700.000 word forms Named entity lexica, compound andamp; tagging rules Finite State Grammers Text Chart Shallow Text Processing Word Level Sentence Level Conceptual System Ontology: Domain-specific semantic knowledge Domain Lexicon: Domain-specific mapping of words to the Conceptual system Tokenizer Lexical Processor POS-Tagger Named Entity Finder Phrase Recognizer Clause Recognizer ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )


Linguistic Analysis and Context Generation : Linguistic Analysis and Context Generation


Slide18 : 1 Ontologies 2 FCA-Merge i) Linguistic Analysis and Context Generation ii) Generating the Pruned Concept Lattice iii) Generating the new Ontology from the Concept Lattice 3 Outlook Agenda


AIFB software : AIFB software uses Propose new concepts/ relations Domain lexicon Text Processing Server uses references Lexical DB


Slide20 : Formal Concept Analysis (FCA) arose in the 1980ies in Darmstadt as a mathematical theory. formalizes the concept of ‚concept‘. is used for deriving conceptual hierarchies from data tables. provides a visualization of the hierarchies by line diagrams. is used in our approach as a method for conceptual clustering.


Slide21 : National Parks in California A formal context about National Parks in California.


Slide22 : Intent B National Parks in California Extent A Def.: A formal concept is a pair (A,B) where A is a set of objects (the extent of the concept), B is a set of attributes (the intent of the concept), AB is a maximal rectangle in the binary relation.


Slide23 : National Parks in California The blue concept is a subconcept of the yellow one, since its extent is contained in the yellow one.


Slide24 : The concept lattice of the context about National Parks


Generating the Pruned Concept Lattice : Generating the Pruned Concept Lattice The ontology concepts are clustered by the algorithm TITANIC.


Slide26 : TITANIC computes the concepts via key sets. Key sets are minimal attribute sets generating a concept intent.


Slide27 : TITANIC computes the concepts via key sets. Key sets are minimal attribute sets generating a concept intent. Key sets form an order ideal in the powerset and can thus be computed à la Apriori  TITANIC. In FCA-Merge, we use the key sets to identify the original concepts of the source ontologies to identify newly generated concepts as suggestions for new concept and relation names


Slide28 : 1 Ontologies 2 FCA-Merge i) Linguistic Analysis and Context Generation ii) Generating the Pruned Concept Lattice iii) Generating the new Ontology from the Concept Lattice 3 Outlook Agenda


Framework : Framework models Domain lexicon Text Processing Server uses references Lexical DB Propose new concepts/ relations uses


Generating the new Ontology from the Concept Lattice : Generating the new Ontology from the Concept Lattice Concepts generating the same formal concept are suggested to be merged. Formal concepts without attributes give rise to new concepts or relations (or subsumptions). Concepts from the same ontology may also be merged. Concepts which generate alone a formal concept are taken over into the new ontology.


Ontology Environment OntoMat : Ontology Environment OntoMat


FCA-Merge (Summary) : FCA-Merge (Summary) Appearance of concepts in documents is discovered. The concepts are clustered. Concepts generating the same cluster are suggested to be merged.


Slide33 : 1 Ontologies and the Semantic Web 2 FCA-Merge 3 Outlook Agenda


Slide34 : Outlook Including axioms and relations in FCA-Merge Structuring queries against the concept lattice Views on (distributed) ontologies Architecture for multiple ontologies (OntoLogging project) Applying FCA-Merge on Semantic Web data  Semantic Web Mining


Slide35 : 1 Ontologies and the Semantic Web 2 FCA-Merge 3 Outlook End


Information Extraction System SMES (DFKI Saarbrücken) : Information Extraction System SMES (DFKI Saarbrücken)