<< Volver atrás

Tesis:

Method for Reusing and Re-engineering Non-ontological Resources for Building Ontologies.

  • Autor: VILLAZON TERRAZAS, Borís Marcelo

  • Título: Method for Reusing and Re-engineering Non-ontological Resources for Building Ontologies.

  • Fecha: 2011

  • Materia: Sin materia definida

  • Escuela: FACULTAD DE INFORMATICA

  • Departamentos: INTELIGENCIA ARTIFICIAL

  • Acceso electrónico: http://oa.upm.es/6338/

  • Director/a 1º: GOMEZ PEREZ, Asunción

  • Resumen: This thesis is focused on the reuse and possible subsequent re-engineering of knowledge resources, as opposed to custom-building new ontologies from scratch. The deep analysis of the state of the art has revealed that there are some methods and tools in the literature for transforming non-ontological resources into ontologies, but with some limitations: Most of the methods presented are based on ad-hoc transformations for the resource type, and the resource implementation. Only a few take advantage of the resource data model, an important artifact for the re-engineering process [GGPSFVT08]. There is no any integrated framework, method or corresponding tool, that considers the resources types, data models and implementations identified in an unified way. With regard to the transformation approach, the majority of the methods perform a TBox transformation, many others perform an ABox transformation and some perform a population. However, no method includes the possibility to perform the three transformation approaches. Regarding to the degree of automation, almost all the methods perform a semi-automatic transformation of the resource. According to the explicitation of the hidden semantics in the relations of the resource components, we can state that the methods that perform a TBox transformation make explicit the semantics in the relations of the resource components. Most of those methods identify subClassOf relations, others identify ad-hoc relations, and some identify partOf relations. However, only a few methods make explicit the three types of relations. With respect to how the methods make explicit the hidden semantics in the relations of the resource terms, we can say that three methods rely on the domain expert for making explicit the semantics, and two rely on an external resource, e.g., DOLCE ontology. Moreover, there are two methods that rely on external resources but not for making explicit the hidden semantics, but for finding out a proper ontology for populating it. According to the provision of the methodological guidelines, almost all the methods provide methodological guidelines for the transformation. However these guidelines are not finely detailed; for instance, they do not provide information about who is in charge of performing a particular activity/task, nor when that activity/task has to be carried out. With regard to the techniques employed, most of the methods do not mention them at all. Only a few methods specify techniques as transformation rules, lexico-syntactic patterns, mapping rules and natural language techniques. In this thesis we have provided a method and its technological support that rely on re-engineering patterns in order to speed up the ontology development process by reusing and re-engineering as much as possible available non-ontological resources. To achieve this overall goal, we have decomposed it in the following objectives: (1) the definition of methodological aspects related with the reuse of non-ontolo-gical resource for building ontologies; (2) the definition of methodological aspects related with the re-engineering of non-ontological resources for building ontologies; (3) the creation of a library of patterns for re-engineering nonontological resources into ontologies; and (4) the development of a software library that implements the suggestions given by the re-engineering patterns. Having in mind these goals, in this chapter we present how the open research problems identified in Chapter 2 are solved by the main thesis contributions. Then, we discuss the verification of our hypotheses, and finally we provide an outlook for the future work in those topics.