A Semantic Approach to Automated Content Augmentation for eLearning Objects ∗

Over the last years IEEE LOM eLearning Objects have been well established as the basic building blocks for educational online content. Equipped with an expressive set of metadata and structured by a variety of named relations, they are nicely suited for self-explorative learning approaches within adaptive hypermedia applications. The authoring of such 'Knowledge Nuggets', though, not only requires content editing, but the provision of meta descriptors and numerous interrelations. Facing the latter in the context of large repositories, where a new object may attain relations to any previously led entity, clearly demonstrates the e ort to be requested from an author. In the present paper we update the status of our educational content management system HyLOs. We introduce instructional design concepts and tools, as well as a content acquisition and analysis toolset, targeting at the semi-automated generation of eLearning Objects. Starting from classroom recordings or o ine content production, automated keyword extraction and classi cation is applied to the raw learning object. In the second part of this paper we rede ne and sharpen the semantic of LOM relations, thereby extending its set by entities missing from the educational perspective. We construct an ontology and inference rules for these inter object relations. Based on this newly introduced Ontological Evaluation Layer and the automated classi cations, appropriate relations between learning objects are autonomously derived. These solutions have been implemented in the Hypermedia Learning Objects System (HyLOs), our prototype of an eLearning content management system. HyLOs is built upon the more general Media Information Repository (MIR) and the MIR adaptive context linking environment (MIRaCLE), its linking extension. MIR is an open system supporting the standard XML, CORBA and JNDI. HyLOs bene ts from manageable information structures, sophisticated access logic and high-level authoring tools like the eLO editor responsible for the semi-manual creation of meta data and WYSIWYG like content editing, allowing for rapid distributed content development.