Enhanced metadata modelling and extraction methods to acquire contextual pedagogical information from e-learning contents for personalised learning systems

Abstract

To make online learning systems (OLSs) effective, it is important to make sure that the learners get the learning objects (LOs) according to their pedagogical suitability and requirements. To assess the suitability of an LO, sufficient information of it is required to be available. These information can be specified as metadata of the document. But there is a dearth of metadata defined for educational documents. Existing standard metadata models like IEEE LOM and others are promising but lack in capturing some crucial learning and pedagogical aspects of LOs. In this paper, we propose a new metadata model that has extended the IEEE LOM to provide an extensive set of metadata for LOs. The proposed metadata seem adequate to describe the contextual learning and pedagogical information of any text and web document based LO. But only specifying the metadata is not sufficient; they need to be extracted from a learning content automatically so that these information can be used by the learners and the OLSs and the educational recommendation systems. Automated extraction of metadata from e-learning contents is a non-trivial task. In view of that, we have provided extraction mechanisms for each of the specified metadata, separately. The experimental results show that the proposed extraction methods are quite accurate in identifying and retrieving the different educational metadata. The statistical inferences of the automated and manual extractions are found to have substantial similarities for each of the extracted metadata element.