Latent Semantic Analysis of Small-Scale Corpora for Positioning in Learning Networks

Positioning in learning networks is the process of establishing a starting point and an efficient route along which a learner may build competences. We investigate computational approaches to services such as positioning that avoid or reduce labor-intense procedures and that are based on the contents of the learning network and the behavior of those participating in it, rather than in predefined procedures and (meta-) data. In this article we consider some preliminary questions related to the use of Latent Semantic Analysis as a computational approach to positioning. In learning networks LSA will often be used on small-scale corpora of learning materials. In this article we develop guidelines for the application of LSA in this type of corpora and we present empirical evidence that substantiates these guidelines. Although the results reported are encouraging we discuss some limitations that need addressing in subsequent work. Introduction Self-organizing learning networks are seen as a solution to the problem of realizing flexible (in time, place and space), personalized (optimal suited to the learner) learning environments for lifelong learning within economical acceptable margins. A learning network is an ensemble of actors, institutions and learning resources which are mutually connected through and supported by information and communication technologies in such a way that the network self-organizes (Koper, Rusman, & Sloep, 2005). In a learning network all actors are involved in furthering the development of competence, which implies that they not only act as learners but 2 LSA of scale small corpora in learning networks will engage in other roles as well, such as consulting expert, teacher, or librarian. Consider, for example, a learning network within the field of veterinary medicine. The learning network contains descriptions of the competences of veterinarians and it offers learning activities and knowledge resources with which learners can develop these competences. Experienced veterinarians may participate to refresh their knowledge and also offer support to students. When a learner (re-)enters a learning network we have to face the 'positioning' problem: where shall the learner be placed in this network and what route through the learning network is the most efficient, considering the competences that the learner wants to develop and the history of the learner. This boils down to identifying the learning activities that need to be completed and those that may be considered redundant considering the needs and prior experience of the learner. Accreditation of prior learning using portfolio assessment and formal assessment …