Dataset-driven Research for Improving TEL Recommender Systems

Abstract

In the world of recommender systems, it is a common prac- tice to use public available data sets from different application environ- ments (e.g. MovieLens, Book-Crossing, or EachMovie) in order to eval- uate recommendation algorithms. These data sets are used as bench- marks to develop new recommendation algorithms and to compare them to other algorithms in given settings. In this paper, we explore data sets that capture learner interactions with tools and resources and that can be used to evaluate and compare the performance of different rec- ommendation algorithms for Technology Enhanced Learning (TEL). We present an experimental comparison of the accuracy of several collabo- rative filtering algorithms applied to these TEL data sets and elaborate on implicit relevance data, such as downloads and tags, that can be used to augment explicit relevance evidence and used to improve the perfor- mance of available recommendation algorithms.