Combining social-based and information-based approaches for personalised recommendation on sequencing learning activities

Abstract

Lifelong learners who select learning activities to attain certain learning goals need to know which are suitable and in which sequence they should be performed. Learners need support in this way-finding process, and we argue that this could be provided by using Personalised Recommender Systems (PRSs). To enable personalisation, collaborative filtering could use information about learners and learning activities, since their alignment contributes to learning efficiency. A model for way-finding presents personalised recommendations in relation to information about learning goals, learning activities and learners. A PRS has been developed according to this model, and recommends to learners the best next learning activities. Both model and system combine social-based (i.e., completion data from other learners) and information-based (i.e., metadata from learner profiles and learning activities) approaches to recommend the best next learning activity to be completed.

Recommending suitable learning paths according to learners’ preferences: Experimental research results

“…In the work, it was found that different approaches to e-learning personalisation were applied so far. Hummel et al (2007) classify personalisation methods into two groups ( Fig. 1): (1) Information-based approachesthey comprise learning technology standardisation, metadata, ontology and application of a semantic web and (2) social-based approaches -usually data mining, collaborative filtering methods, etc.…”

Report of the Results of an IMS Learning Design Expert Workshop

“…The two layers are then connected via a mechanism that feeds most needed and most popular changes from the feedback data layer back into the LD-templates of the first layer. Hybridly combining fixed ontologies with flexible social behavior (this is the same as the two-layered design structure) has recently been successfully applied in providing personalized recommendations in learning networks [7]. Therefore, the setup of the proposed solution has been justified.…”

“…Like described by Hummel et al [7], we used indirect social navigation and collaborative filtering (data mining) techniques to derive the advice. When most peers having the same or similar user profile would have successfully completed B after having completed A (data stored in a transition matrix), it would be most likely that A, B, ... would become the 'standardized' sequence for these students.…”

A contextual semantic representation of learning assets in online communities of practice

“…mainly related to finding the adequate learning assets (Tzikopoulos et al, 2007;Davies et al, 2008;Allen and Mugisa et al, 2010;Manouselis et al, 2011), to the quality of content (Tzikopoulos et al 2007;Assche et al, 2009) and on the way in which this content is delivered to the learner (Hummel et al, 2007;Janssen et al, 2011).…”