Issues and Considerations regarding Sharable Data Sets for Recommender Systems in Technology Enhanced Learning

Drachsler, H., Bogers, T., Vuorikari, R., Verbert, K., Duval, E., Manouselis, N., Beham, G., Lindstaedt, S., Stern, H., Friedrich, M., & Wolpers, M. (2010, 28 September). Issues and Considerations regarding Sharable Data Sets for Recommender Systems in Technology Enhanced Learning. Presentation at the 1st Workshop Recommender Systems in Technology Enhanced Learning (RecSysTEL) in conjunction with 5th European Conference on Technology Enhanced Learning (EC-TEL 2010): Sustaining TEL: From Innovation to Learning and Practice, Barcelona, Spain.The presentation is based on the positioning paper of the dataTEL Theme Team of the STELLAR Network of Excellence (http://www.teleurope.eu/pg/groups/9405/datatel/) that addresses the lack of educational data sets in TEL and present ideas to overcome this situation. The accompanying paper: Issues and Considerations regarding Sharable Data Sets for Recommender Systems in Technology Enhanced Learning, can be found at http://www.sciencedirect.com/science/journal/18770509 and a pre-print is available in our Dspace repository and at scribd. The presentation starts with a description of the current situation where almost none educational data sets are publicly available. This is a strange situation as plenty of data is saved on a daily base in LMS like Moodle, Blackboard. In other domains like e-commerce it is a common practice to use publicly available data sets from different application environments (e.g. Yahoo, MovieLens) in order to evaluate algorithms and create new data products. These data sets are for instance used as benchmarks to develop new recommendation algorithms and compare them to other algorithms in certain settings. Recommender systems are also increasingly applied in Technology Enhanced Learning field but it is still an application area that lacks such publicly available data sets. Although there is a lot of research conducted on recommender systems in TEL, they lack data sets that would allow the experimental evaluation of the performance of different recommendation algorithms using comparable, interoperable, and reusable data sets. This leads to awkward experimentation and testing such as using data sets from movies in order to evaluate educational recommendation algorithms.Stella