Learning Objects Reusability and Retrieval through Ontological Sharing: A Hybrid Unsupervised Data Mining Approach

摘要

Ontologies add semantics and context to learning objects (LOs), enabling LO sharing and reuse in a contextual learning environment and providing better navigation and retrieval of LOs. However, the effectiveness of LO reuse from LO repositories is compromised due to the use of different ontological schemes in each LO repository. This paper presents an algorithmic framework for ontology mapping and merging, OntoDNA, which employs hybrid unsupervised data mining techniques to resolve the semantic and structural differences between ontologies to subsequently create a merged ontology to facilitate LO reuse and retrieval from the Web or from different LO repositories such as ARIADNE, MERLOT, CAREO or Educause. Experimental results on several real ontologies and comparisons with other ontology mapping and merging tools demonstrate the viability of the OntoDNA in terms of precision, recall and f-measure to interoperate LOs in the LO repositories.