Article Preview
TopIntroduction
Research on Ontology Engineering methodologies has provided methods and techniques for developing ontologies from scratch. Well-recognized methodological approaches such as METHONTOLOGY (Gómez-Pérez, Fernández-López & Corcho, 2003), On-To-Knowledge (Staab, Schnurr, Studer & Sure, 2001), and DILIGENT (Pinto, Tempich & Staab, 2004) give guidelines that help researchers to develop ontologies. However, researchers face an important limitation: no guidelines are provided for building ontologies by re-engineering existing knowledge resources widely used by a particular community.
During the last decade, specific methods, techniques and tools were proposed for building ontologies from existing knowledge resources. First, ontology learning methods and tools have been proposed to extract relevant concepts and relations from structured, semi-structured, and non-structured resources (Gómez-Pérez & Manzano-Macho, 2004; Maedche & Staab, 2001) in order to form a single ontology. One important constraint to these methods and tools is that they propose ad-hoc solutions to transforming such resources, mainly texts, into ontologies. Hepp et al. (Hepp, 2006; Hepp & Brujin, 2007; Hepp, 2007) stated that employing methods and techniques when ontologizing non-ontological resources to the level of ontologies is key for the success of semantic technology for two main reasons: (1) if the use of semantic technologies for real-world data integration challenges is required, it is possible to refer to the original conceptual elements, and (2) for many domains, the existing category systems, XML schemas, and normative entity identifiers are the most efficient resources for engineering ontologies.
The ontologization of non-ontological resources has led to the design of several specific methods, techniques and tools (Hepp & Brujin, 2007; Hyvönen, Viljanen, Tuominen & Seppälä, 2008; Gangemi, Guarino, Masolo & Oltramari, 2003; García & Celma, 2005). These are mainly specific to a particular resource type, or to a particular resource implementation. Thus, everytime ontology engineers face a new resource type or implementation, they develop ad-hoc solutions for transforming such resource into a single ontology.
In parallel, and within the context of the NeOn project1, a novel scenario-based methodology for building ontology networks2 has been proposed: the NeOn Methodology (Suárez-Figueroa, 2010; Gómez-Pérez & Suárez-Figueroa, 2009). One of these novel scenarios is Building Ontology Networks by Reusing and Re-engineering Non-Ontological Resources. As opposed to custom-building silos of single ontologies from scratch, this new scenario emphasizes the re-engineering of knowledge resources for building ontologies that are connected with other ontologies in the ontology network.
The motivation of this paper lies in this scenario of the NeOn Methodology and the use of re-engineering patterns to transform the non-ontological resources components into ontology representational primitives. Along this paper we will try to demonstrate that the use of re-engineering patterns for transforming non-ontological resources into ontologies has several advantages: (1) they embody expertise about how to guide a re-engineering process, (2) they improve the efficiency of the re-engineering process, and (3) they make the transformation process easier for ontology engineers.