Feasibility Study of Recommender Systems in Academia

Image courtesy of Laia Ros, Flickr

Recommender Systems (RSs) are systems capable of predicting the preferences of users over sets of items (given the historical user-preference data). RSs can be found almost everywhere in the digital space (e.g. Amazon, Google, Netflix), shaping the choices we make, the products we buy, the books we read, or movies we watch. However, there are almost no RSs in the academic world, where we expect they can have a great potential.

One of the rare exceptions in this context is the Master Orientation Tool, operating at University College Maastricht since 2012. The bachelor's students from this study programme use this recommender system to discover master's programmes that fit their past, current, and future courses. The system allows the students to modify their own course choices to explore study alternatives and how they influence their master's programme possibilities.

The success and popularity of the Master Recommendation System demonstrates the importance of RSs in academia. We foresee two levels of application of RSs in the university context: internal and external. On the internal level, RSs can help students to choose courses, teachers, academic programs, thesis topics, internships, employers, jobs, etc. On the external level, RSs can help prospective students, parents, internship providers and future employers to better match preferences over the choices they have related to the university. On both levels providing the right information will cause better adaptation of universities in a long run. Thus, we may conclude that RSs are not just applicable but urgently needed for modern universities. To address these issues, we define the main goal of this WUN project proposal as to provide a feasibility study of applicability of RSs in the university context.

Broad aims of the project:

  • Defining recommendation tasks relevant to the university context.
  • Overview of RSs approaches suitable for these tasks (e.g. content-based, collaborative filtering, hybrid, reciprocal).
  • Developing a methodology for implementing university RSs (conceptualization, formalization, data gathering, RSs tool selection, evaluation).
  • Overview and evaluation of the suitability of existing RSs tools (freely available or commercial).
  • Identifying international initiatives that can help developing RSs for WUN universities.

To implement the project, we propose to form a collaborative network of three groups from WUN Universities that have experience in RSs. The main result we expect is a report “Guide to Building Recommender Systems for Academia” that will address the five tasks described above. This document will be available to all the WUN partners and can be used for setting up new university RSs. Relevant parts of the report will be submitted as separate joint publications of the research team (international conferences/journals). In addition, a website with up-to-date links to RSs tools will be developed.

The project will be implemented in stages via online meetings, faculty exchanges, and workshops. In addition to the three WUN partners, the Open University (The Netherlands) and Data-Science Consultancy (The Netherlands) also cooperate in the project.

  • Dr. Evgueni Smirnov, Maastricht University
  • Dr. Kurt Driessens, Maastricht University
  • Dr. Irena Koprinska, University of Sydney
  • Dr. Kalina Yacef, University of Sydney
  • Professor Osmar Zaïane, University of Alberta

New members in 2016:

  • Dr. Gerasimos Spanakis, Maastricht University
  • Dr. Stylianos Asteriadis, Maastricht University
  • Dr. Simon Price, Advanced Computing Research Centre, University of Bristol