Abstract:

This paper describes a platform for distance learning that can adapt course contents to the learning preferences of each student.

The existing learning management systems (LMS) lack interactivity and pedagogy, leading to learning models that depend largely on student motivation. Artificial Intelligence techniques such as Case-Based Reasoning (CBR) and Fuzzy Systems (FS) are used to overcome these problems as a basis for building an Intelligent Tutoring System (ITS) that is able to replace, to some extent, a human teacher.

Various models of learning styles have been proposed to determine the learning preferences of a student. These preferences may be related to the way students prefer to receive information. There are different learning style models such as Felder-Silverman model that classifies the learning style of a student into 4 dimensions: Sensory / Intuitive, Visual / Verbal, Active / Reflective, Sequential / Global.

Besides this, contents classification is necessary in order to relate a learning object to each student cognitive level. Bloom taxonomy is used to achieve this.

The VARK paradigm also allows to determinate the learning preference of an individual. The term learning preference refers to the way that a person has a greater willingness to receive the information. VARK considers four different types of learning: (V) visual, (A) aural, (R) read/write and (K) kinesthetic. Each course contents is classified according to Bloom and VARK.

The learning profile of each student is initially determined by a Felder-Silverman questionnaire and then mapped into VARK preferences by means of fuzzy logic. This mapping is based on the Mamdani inference.

In ADAPT, each learning object (LO) is classified according to Bloom Taxonomies and VARK dimensions. This way, the same contents may be presented in various ways, “more” in a visual, aural, read & write or kinestekic way. To support this, the allowed file types are text, pdf, video, slide, sound, external link, test, flash, applet and image, all according to the SCORM standard.

But, besides this, the path that each student follows along the course, is created according to the type of pages that more closely match his learning style. The support structure for a course script is a graph, distinct for each student. As each page he is allowed to visit is cataloged according to VARK and Bloom taxonomy, these pages are presented according to the best match between VARK classification / student learning profile and Bloom Taxonomy / student learning level.

The outcomes of each student at each course are stored by the system. Together with course scripts and learning style, each one is an example of success or failure, and constitute a Case-Based reasoning library (CBR). For future students the system matches learning profiles with these past cases and gets the more appropriate script for each student.

Optimizations may have to be done if the learning profile changes along the course. The changes are detected by data-mining techniques and optimizations are carried out by genetic algorithms, if needed.

The ADAPT LMS has been preliminary tested for teaching Digital Systems at ISEC – Instituto Superior de Engenharia de Coimbra / Instituto Politécnico de Coimbra. The project is founded by FCT - Fundação para a Ciência e Tecnologia - PTDC/CPE-CED/115175/2009 and FEDER - Eixo I of Programa Operacional Factores de Competitividade (POFC) / QREN (COMPETE: FCOMP-01-0124-FEDER-014418).

Keywords:

e-learning, CBR, Fuzzy Logic, ITS, Adaptive learning systems, intelligent systems.