Link Recommendation in E-learning Systems Based on Content-Based Student Profiles

DANIELA GODOY; ANALIA ADRIANA AMANDI

Handbook of Educational Data Mining

CRC Press

Lugar: Boca Raton; Año: 2010; p. 273 - 286

E-learning systems offer students an opportunity to engage in an interactive learning processin which they can interact with each other, teachers, and the learning material. Mostof these systems allow students to navigate the available material organized based on thecriteria of teachers and, possibly, the student learning style, and background. However, theWeb is an immense source of information that can be used to enrich the learning processof students and, thus, expand the imparted knowledge and acquired skills provided by theE-learning systems. In order to exploit the information available on the Web in a fruitfulway, material should be carefully selected and presented to students in the proper time notto overload them with irrelevant information.Many approaches have emerged in the past few years taking advantage of data miningand recommendation technologies to suggest relevant material to students according totheir needs, preferences, or behaviors. These techniques analyze logs of student behaviors,their individual characteristics, behavior and learning styles, and other data to discoverpatterns of behavior regarding content in the system that are later applied to personalizethe student?s interaction with the system. Learning of association rules, induction of classifiers,and data clustering have been used in several works to enhance the effectiveness inthe presentation and navigation of content in E-learning systems, consequently enrichingthe learning experience of students.In this chapter, we present a recommendation approach for suggesting relevant learning material to students. This approach aims at acquiring comprehensible content-based profiles capturing student interests starting from observation of their behavior in an E-learning system. These profiles are conceptual representations of the material read by the students as they take the courses offered in the system. Not only long-term interests represented in the profiles are considered to make recommendations in this approach, but also the context in which the student is acting at a certain moment. To accomplish this goal,the context of student activities is described using his or her profile and novel information is gathered from the Web matching their active interests.The rest of the chapter is organized as follows. Section 19.2 discusses related works on recommendation approaches used in the context of E-learning systems. Section 19.3 describes the proposed approach for recommendation and its integration with user profiling.Both the construction of content-based profiles and the detection of the active interestsof a student are described in Section 19.3, which also introduces the mechanisms to generate and deliver recommendations. Section 19.4 shows the results obtained in a course. Finally, concluding remarks are stated in Section 19.5.