ON ADAPTING LMS FOR RECOMMENDATION AND PERSONALIZATION BASED ON CONTEXT-AWARE TECHNOLOGIES
Learning Management System (LMS) has consolidated itself as a flexible and dynamic environment for supporting content management and academic processes to both on-site and online students and lectures (Maldonado et al., 2011). Despite all advantages, some studies show that there are still several challenges to face before building complete and successful LMS environments. These challenges emerge mainly due to lack of knowledge about experiences and predilections of the myriad of students and lecturers that use LMSs with different abilities, preferences and working style (Bhuasiri et al., 2012).
As a result of the aforementioned studies, two aspects are highlighted as the most immediate and affordable when improving an LMS: personalization and recommendation. On the first hand, personalization in LMS is focused on the adaptation of learning resources and services according to the students' real needs (Peter et al., 2010). There are different (but non-exclusive) alternatives to achieve this personalization: content adaptation, browsing-centered, customized interfaces (especially for disabled people), and device-dependent interfaces. On the other hand, an LMS should also show a proactive behavior, namely to be able to recommend contents and services both to students --in their learning process-- and lecturers --in their decisions on the course management (Silva et al., 2012). As an example, an LMS with personalization and recommendation capabilities could be very useful when a student receives poor grades in a subject. In that case, the LMS could offer personalized assignments to help her in identifying her mistakes and recommend new contents to reinforce her knowledge.
One of the main alternatives to cope with the augmentation of LMS with personalization and recommendation services is based on context modeling by means of ontologies (Nganji et al., 2011). They enable formal and shared descriptions of terms related to LMS, which can therefore be processed and exchanged by software applications. Moreover, they allow the extraction of implicit knowledge derived from the classification and restrictions modeled in the ontology. All in all, ontologies are essential in a scenario where the semantic information should be taken into account. As the LMS records a large volume of data and events generated by the users, the semantics of such data and events could be captured and used as input in a context-aware system integrated into any LMS.
In this paper we follow the research line based on context modeling through ontologies as a solution to offer personalization and recommendation in LMS. In particular, we first study what tools are the most frequently used in these environments and what events are usually generated by these tools. Then we define the requirements of an LMS ontology based on these indicators and the way to combine it with a profile ontology and a competence evaluation ontology from our previous work (Cantabella et al., 2013). As a result, we will obtain a knowledge model able to represent semantic information about LMS components, students' and lecturers' profiles and academic results. As a proof of concept, we will integrate our context-aware system into Sakai (http://www.sakaiproject.org/), a well-known LMS which is currently being used in our University.