RECOMMENDATION SYSTEMS USING LEARNING ANALYTICS FOR CURRICULUM DEVELOPMENT AND STUDENTS PROGRESS OPTIMIZATION

As a „university desktop“ the student life cylce management system provides self services and administration applications regarding curriculum design, course delivery and exam records. Build upon curriculum data of students progress a web-based recommendation application was developed to support information and selection of teaching modules in study programs with partly elective characteristics. It features different recommendation algorithms that are based on content-based, collaborative and conversational recommendation approaches.
For example, collaborative filtering uses the records of similar students in order to recommend modules and subjects to students while content-based and conversational recommendation techniques allow students to interact with the module selection. In addition, lecturers can assign approved prerequisite options for recommending new modules.

Thus the application provides searching and browsing for modules and ranks results by different metrics. For matching individual needs and interests, users are able to select the most interesting metrics and to apply several filters for different outcomes. This advantage comes along with attendance figures on modules typed as elective. Students can see how many other students choose a specific elective field and lecturers get information on attendance of their courses.

Such information can be also useful for academic planning of courses as well as further development of degree programs. With the integration of learning analytics, the experiences of the current cohort can be applied to future remodeling of a study program. It also enables the capacity planning to take into account students who take exams in advance or who re-sit exams in a later semester. Semantic analysis of module descriptions on the other hand can be used for the classification of learning objectives to support competence-oriented examinations.

In a two-tier evaluation, students from different degree programs are first asked to measure the quality of the module recommendations as well as an grade prediction function in terms of usefulness. In the further course of the study, students progress paths and module choice motives are investigated in more detail. This analysis will help students to find a suitable path through complex program structures and with the semester organization.