ConcepTool uses different kinds of reasoning services at the conceptual level to perform a detailed analysis of the domain knowledge base that is being built, modified or combined. Analysis checks the consistency of this knowledge and highlights its implications, i.e. further relationships between concepts derived by their respective structures and by their mutual constraints. The reasoning services used for analysis provide deductions based on Description Logics, linguistic inferences based on lexical ontologies, or heuristic inferences based on structural rules.

ConcepTool can model UML, OKBC, or DL classes, UML associations, ER relationships, and DFD or IDEF processes using a frame-based knowledge representation. Each of these concepts can include own and template slots with facets (e.g. attributes), partonomic and taxonomic relationships, and global constraints (e.g. disjointness or coverage).

ConcepTool supports the virtual combination of different domain knowledge bases and application ontologies, creating a shared view of their common components called articulation. Potentially related concepts are proposed using the available reasoning services and detailed mappings are automatically introduced to relate source conceptsto their images in the articulation.

ConcepTool fact-file

What's the problem?

• Most systems and tools that support the modelling and the combination of domain knowledge and ontologies have limited reasoning functionalities to analyse this knowledge. Some of these systems and tools have no reasoning functionalities at all.

• Available reasoning functionalities in domain knowledge modelling systems generally belong to one of three different kinds: logic deductions, heuristic inferences, or lexical inferences. There is no "fusion" mechanism to dynamically combine all these different inferences, which would improve the analysis (and thus the sharing and the reuse) of domain knowledge.

• The inferential engines and algorithms that are currently used to provide logic-based deductive services do not operate at the conceptual level. In fact, they fail to distinguish between different deductive mechanisms for diverse categories of concepts. For instance, the differences between hierarchies of classes and hierarchies of functions are not taken into account.

Towards a solution

Domain knowledge modelling in ConcepTool is performed at the conceptual level, while modelling in most of the other existing tools is performed at the epistemological level. In other words, the underlying knowledge model of ConcepTool:

• explicitly distinguishes between different categories of concepts (e.g. classes, relationships, functions, goals).

• integrates semantic constructors from frame-based, DL-based, conceptual, and lexical models.

Domain knowledge analysis in ConcepTool is automatically performed by using a wide spectrum of deductive (logical) and inferential (heuristic) services of different kind.  Systems like OILed, OntoSaurus and OpenKnoMe, which are based on Description Logics engines, only compute subsumption between standard ``class'' concepts. Conversely, ConcepTool combines various inference mechanisms (e.g. heuristic inferences and logical deductions), providing different kinds of reasoning functionalities (e.g. structural subsumption between associations, heuristic relationships between attributes, lexical equivalence between classes). The rationale for a wider number of deductive behaviours is twofold.

• The different categories of elements in ConcepTool have different structures and different kinds of properties. Therefore, all these heterogeneous conceptual elements must be analysed using different reasoning rules (such as different hierarchy-forming rules).

• Reasoning at different levels (structural, lexical, heuristic) helps to highlight all the possible implications of knowledge (such as deriving further hierarchical relationships between domain concepts). In fact, structural reasoning (e.g. subsumption between two classes) only establishes an objective link between objects which share the same properties (e.g. attributes). Conversely, lexical and heuristic reasoning can suggest further links between apparently unrelated objects or attributes in the domain.

Domain knowledge sharing and reuse in ConcepTool is performed through the combination of different techniques from both ontology-oriented merging and alignment and database-oriented information integration. By providing reasoning services which identify and validate mappings between similar concepts in distinct DKBs, ConcepTool supports the creation of a virtual shared view (the articulation). This can be used either to query distinct DKBs in a distributed context or as a starting point in their full integration.

ConcepTool has been explicitly conceived as a support system for combining domain knowledge, i.e. for aligning, articulating, integrating, and reusing ontologies and DKBs. Therefore, reasoning in our system includes features which are absent in other similar tools. For instance, parts of the expressive power of a DKB under analysis can be selectively excluded for the purpose of reasoning. In this way, the quantity and the quality of logic deductions can be modulated. Selective exclusion generates different deductive results which could be useful to highlight further links between pairs of concepts, one in each of the DKBs to be combined.

Take a Guided Tour

(Forthcoming)

Try a Demonstration

You can download ConcepTool from : http://www.csd.abdn.ac.uk/research/IKM/ConcepTool

Technical requirements: Windows 98, 2000, XP, Java Runtime Environment 1.3 Windows & Unix multiplatform version; only tested & guaranteed under Windows at the moment)

Example Applications

AKT Articulation Framework

AKT Ontology Management

Further Reading

Key document:

H. Meisel and  E. Compatangelo (2002) EER-ConcepTool: a "reasonable" environment for schema and ontology sharing. 14th IEEE International Conference on Tools with Artificial Intelligence (ICTAI'2002).

 Other relevant documents:

Compatangelo, Ernesto and Meisel, Helmut (2002) K-ShaRe: an architecture for sharing heterogeneous conceptualisations. 6th International Conference on Knowledge-Based Intelligent Information & Engineering Systems.

Compatangelo, Ernesto and Meisel, Helmut (2002) Intelligent support to knowledge sharing through the articulation of class schemas. 6th International Conference on Knowledge-Based Intelligent Information & Engineering Systems.

E. Compatangelo and H. Meisel (2002) EER-CONCEPTOOL: conceptual analysis of EER schemas and ontologies. Technical Report, Dept. of Computing Science, University of Aberdeen.

Semantic representation

View in the AKT Triplestore Browser or as RDF.

Also available in DOAP RDF (Description Of A Project)