Ontology-driven unification of conceptual data modelling languages


Unificación de Lenguajes de Modelado Conceptual de Datos dirigidos por Ontologías


Introduction and objectives Descripción y objetivos - participants partecipantes - outputs resultados

Project funded by the South African Department of Science and Technology and the Argentinian Ministerio de Ciencia, Tecnología, e Innovación Productiva, 2012-2015 (3 years)

Overview

Introduction and background

There has been a growing interest in the development and use of domain ontologies, strongly motivated by the Semantic Web initiative. Such domain ontologies aim at the construction of computer-processable logical theories that represent some part of the real world. This overlaps with the established field of conceptual data modelling, which aims at providing models for software applications. Both play an important role in several areas.

Artificial intelligence turns out to require the representation of a lot of human knowledge in order to act "intelligently". To achieve this, they rely on conceptual models using knowledge representation languages, such as semantic networks and the recently W3C standardised Web ontology language OWL, which is based on Description Logics, of which its main strength is its infrastructure for automated reasoning.

Within conceptual modelling, one can identify, among others, Object Role Modelling (ORM), the Unified Modeling Langauge (UML), and Extended Entity-Relationship diagrams (EER). ORM is a fact-oriented modeling approach for specifying and transforming information at the conceptual level, which relies heavily on natural language, has a very expressive and intuitive diagram language, and uses sample populations. Entity-Relationship (ER) modelling is the traditional method and language of choice for relational database design. Within object-oriented software development, the community has also proposed viewing software components (classes/objects) as models of real-world entities, using UML. UML, accepted by the Object Management Group, is the de-facto standard language in software engineering, since its approach facilitates the transition to object-oriented code. UML includes class diagrams that may be annotated with expressions in a textual constraint language as well as formally with the Object Constraint Language (OCL).

These usages and approaches to ontologies and conceptual models entail the use of some language, which influences the kinds of details that can be considered to model. Also, all languages and associated methodologies have tools that support the engineer in the task of creating, validating, and updating models in that specific language. These tools help in creation of large models and there are proof-of-concept tools to integrate them, but they are limited to only those represented in the same language. With company mergers, upscaling of scientific collaboration in the life sciences, and the emergence of the next generation Internet---the Semantic Web---integration among different sources has become an imperative. The subtle representational and expressive differences in the languages makes this task very difficult, and current tools offer only very limited functionality in importing and exporting partial models into other languages. In addition, the strengths of each language are limited to that language only, mainly due to historical reasons.

Aims

We aim at developing the formal conceptual apparatus and tools to integrate heterogeneous conceptual models, based on clear semantic specifications and with tractable algorithms to ensure implementability and usability. This will enable (i) the creation, editing, managing, and storing of several interconnected existing ontologies and conceptual data models, (ii) the cross-fertilization of each language's strength to the other.

Specific objectives

The specific objectives of the project are threefold:

Resumen

TBA

Palabras claves: Representación del conocimiento, Ingeniería de ontologias, Web Semántica, Modelado conceptual, Lógicas para la Descripción

Participants partecipantes


Outputs resultados