Most problems facing Distance Education (DE) academic advising can be overcome using a course recommender system. Such a system can overcome the problem of students who do not know their interest in courses from merely their titles or descriptions provided in course catalogues. The authors introduce in this chapter an XML user-based Collaborative Filtering (CF) system called CRS. The system aims at predicting a DE student's academic performance and interest on a course based on a collection of profiles of students who have similar interests and academic performance in prior courses. The system advises a student to take courses that were taken successfully by students who have the same interests and academic performance as the active student. The framework of CRS identifies a set of course features for every academic major. The authors experimentally evaluate CRS. Results show marked improvement.
TopIntroduction
Currently, Recommender Systems (RS) are widely used in different domains such as E-commerce and digital libraries. In general, recommendation systems suggest items or products, by analyzing what users with similar tastes have chosen in the past. Recently, recommender systems are applied in the e-learning domain in order to personalize learning content. In this process, students are matched with the appropriate learning objects. Students are connected with each other according to their individual interests, skills, needs, and learning goals. A number of research works advocate the use of recommender system in e-learning systems and learning management systems. Calvo (2003) uses recommender system in intelligent learning system. Andronico (2003) suggests educational resources to students through mobile phones. Liu & Gree (2005) provides a framework that selects a list of learning objects that suit students.
There are a number of approaches for recommending learning objects. One of these approaches is Content-Based System (CBS) recommend courses to students based on the content of courses and the student’s preferences. The approach detects the similarities between courses attributes (such as name, abstract, keywords) and other courses. The student enters a course’s attributes and the system recommends courses that have similarities with the active course’s attributes. Another approach is Rule Based Filtering. The approach filters courses based on a set of rules to be applied to the student’s profile and the system’s profile.
There have been a number of works that have addressed on-line automatic advising and predicting student performance in e-learning (Vance, 2004; Khribi, 2007; Thai-Nghe, 2010). Mohamed (2007) provides techniques for on-line automatic recommendations in e-learning systems using the access history of learners. The work of (Vance, 2004) provides a guide to developing e-advising standards for advisees, advisors, and administrators. The work of (Thai-Nghe, 2010) uses recommender system techniques for educational data mining and for predicting student performance.
A DE student may not be able to know his interest in a course from merely its title or from the description of the course provided in the course catalogue. Also, the advisor needs to advise the student to take a course that suits the student’s academic performance and skills. Towards this, the advisor needs to consider the performance of students in all his prior courses, which is time consuming. These problems can be overcome using a course recommender system. We introduce in this chapter a type of Collaborative Filtering (CF) system called Course Recommender System (CRS). The system aims at predicting a student’s academic performance and interest on a course based on a collection of profiles of students who have similar interests and academic performance on prior courses.