Aplicaciones de la expansión de consultas basadas en ontologías de dominio a la búsqueda de objetos de aprendizaje en repositorios

Abstract

In the e-learning field, great efforts are made to improve the learning process. One of these is focused on motivating the use and the reuse of digital resources in repositories. Initially, efforts were focused on increasing the amount of available resources. Nowadays, there are a lot of resources stored in repositories so the challenge has become that of improving the search, selection and access to resources scattered and distributed throughout the repositories. In this sense, the objective of this thesis is to propose a strategy for query expansion based on domain ontologies which allow the instructional designer to find objects in a repository which are relevant for the design of their courses or to compose other, more complex resources. To achieve this goal, we analyse the proposals of query expansion either in the field of information retrieval in general, or specifically on learning object repositories. Based on the aforementioned, the criteria for query expansion based on ontologies have been established, and the ways to resolve the problems identified have been defined. Finally we propose, design and implement the query expansion strategy based on ontology domain applied in the context of the search for learning objects in repositories. For our proposal, an experiment in the area of genetics has been designed using the Gene Ontology as a knowledge base and the MERLOT repository as a provider of learning objects in this domain. The test queries are defined using the contents dealt with in a set of courses in genetics published on the Web by higher education institutions in 2009. The evaluation of the results is carried out by three experts in the field. The analysis of the correlation and association between the experts' evaluations is carried out through the analysis of Cohens' Kappa, and Spearman's correlation coefficient. Finally, the effectiveness of the query expansion proposal is evaluated based on the metrics of coverage and novelty applied to the retrieved results of queries, with and without expansion. The main contribution to this research is a strategy for query expansion based on domain ontologies to allow the instructional designer to obtain relevant results that, without the expansion, could not be retrieved by the repositories of learning objects. We expect that if the instructional designer can access relevant resources, this may contribute to the quality of e-learning courses or the quality of the new resources created from them. It is worth noting that the effectiveness of our proposal is affected by the information retrieval system used in each repository, the quality of stored resources and their labelling, and the completeness and quality of the knowledge base used for expansion.