Mecanismos semánticos orientados a la flexibilidad para los repositorios de objetos de aprendizaje

Abstract

Current approaches towards enhancing reusability of learning materials in digital format use the concept of learning object as the key element of a new approach based on the distributed availability of learning resources. This model relies on the existence of learning object repositories creation, such as MERLOT or CAREO, whose main purpose is to classify different Web resources through metadata descriptions. However, the existence of multiple learning object definitions point out the need of a new generation of flexible learning object repositories capable to fit any existent conceptualization of the term. In this scenario, ontologies play an important role as the basis to provide a sound semantic model that fulfills the new requirements. Nonetheless, flexibility is not the only lack in these repositories. The lack of conformance with current specifications and standards has also been marked as an issue and thus analyzed in the present research. The learning object conceptualization, as described by current standards and specifications such as IEEE LOM and ADL SCORM, is not the one used in the most repositories. This fact hampers the metainformation management process associated to the use of learning objects. Today, most applications do not use any formal model of commonsense knowledge. Advanced knowledge bases such as Opencyc make it possible to create semantic relationships between the concepts defined, which allows concepts such as country, king, doctor or relationships like X near Y to be linked without ambiguities nor changes due to possible interpretations. This fact, if applied to learning object repositories, provides new inference possibilities over learning object metadata records and allows meaningful descriptions of the metainformation stored in those repositories, as it turns text-based metadata records into machine-understandable semantic metadata records. This dissertation proposes the use of this kind of knowledge through the definition of a semantic schema aimed at allowing the meaningful description of the information in existing metadata records repositories. To that end, it will be necessary to define the complete architecture of what it will be called a semantic learning object repository, based on a formal model described in an ontology language (OWL in particular). This architecture will facilitate external software applications or agents the metainformation management tasks. Running inferences on the knowledge stored inside the repositories’records, will extend current repositories’functionalities, introducing key advantages and features.