Article Preview
Top1. Introduction
Conceptual models represent the static characteristics of a system, and they refer to models that have been formed after they have been conceptualized. Ontology is a formal description of a domain conceptualization (Proper & Guizzardi, 2021), which is composed of a set of names for concepts, roles and individuals to represent a knowledge base composed of a terminology set and a set of assertions. In the terminology part (TBOX), we define axioms for general concepts and role inclusions. The roles (relations) can be between pairs of concepts (called object properties) or between concepts and data types (called data type properties). The set of assertions (ABOX) are instances of concepts (concept membership axioms) to assert that an individual (object) belongs to a concept, instances of object properties (object property membership axioms) to assert that an individual has a relation to another individual, or instances of data type properties (data type property membership axioms) to assert that an individual has a data property.
The main domain of ontology application is the Semantic Web, where Web documents are annotated with information (meta-data) from ontology terminology. These Web documents are instances of ontology concepts and roles. Thus, the annotation process is considered as the creation of assertions. The terminology and assertion sets represent the knowledge base of Web agents (programs). The Semantic Web has an architecture composed of resources (Web documents) and a set of intelligent agents. Each one has its own ontology.
As an alternative to keyword matching, ontologies are used in information retrieval as an intelligent search tool via an inference mechanism. The role of these intelligent agents is to answer user queries by running inference rules on their own knowledge base (ontology). Web intelligent agents can derive conclusions from shared knowledge bases linked by URI (Universal Resource Identifier) references. If an intelligent agent finds in its local query answer a reference to a distant knowledge base, it sends the URI reference to the appropriate agent. The former will dereference the URI reference to find its ontology and infer from its knowledge base to send back the result as a response. In this way, all the Web intelligent agents communicate to answer user queries using their proper ontologies (Bourahla, 2018).
An ontology describes Web resources as a graph-based model and it is formalized in the language OWL (Web Ontology Language), which is based on RDF (Resource Description Framework). The graph-based model can be serialized into a set of triples using one of several syntaxes, including XML, Turtle, and functional formats. A triple is composed of a subject, a predicate and an object. The subject is a Web resource (document). The object can be a Web resource (document) or a literal. The predicate is an OWL or RDF (RDFS) property (relation) that relates the subject to the object. The OWL/RDF model contains the ontology terminology and assertions as a set of RDF triples. Thus, with this OWL/RDF model, we can declare axioms, like concept inclusion axioms and membership axioms.
There are many technologies for developing ontologies that propose methodologies for developing ontology conceptualization, such as Ontolingua (Fikes et al., 1997), which was developed by Stanford University's Knowledge System Laboratory. Ontolingua is devoted to ontology development using a form-based Web interface, and tools to formalize the conceptualization results, like Protégé (Tudorache et al., 2011).
In this paper, we propose a tool for modeling and formalization of ontologies. A graphical editor was developed to model ontology conceptualization with respect to the defined meta-model in the Ecore standard, using UML notations. Ecore (Steinberg et al., 2008, 2009) is the core meta-model of EMF. It allows the expression of other models by leveraging its constructs. The environment is defined in terms of itself (its own meta-model). The modeling result will be transformed using Triple Graph Grammars (Anjorin et al., 2015) into an ontology model with respect to a defined OWL2 ontology meta-model developed with Ecore. This bidirectional transformation can be realized by executing a set of rules defined with respect to a correspondence meta-model between the ontology-conceptual meta-model and the OWL2-based ontology meta-model.