Martijn Willemsen The recommender LAB of Martijn Willemsen researches how decisions can be supported by recommender systems, and includes domains such as movies, music, health related decisions and energy-saving measures. The LAB focuses on how insights from decision psychology can improve recommender algorithms, how to best evaluate recommender systems and on novel recommendation methods that help users with developing preferences and goals.


Recommender systems are ubiquitous in our current information-driven society. Without recommender algorithms, we would face major difficulties finding relevant content on facebook or twitter, would spend a lot of time discovering relevant movies to watch or music to play and might get lost in online shopping environments that feature thousands of similar products. The area of recommender systems has developed as a multidisciplinary field including researchers in computer science developing novel algorithms and new machine learning and deep learning methods to improve recommendations, researchers in HCI trying to improve the user experience of these systems, researchers in marketing and e-commerce studying the commercial value of these systems and researchers in decision psychology trying to improve algorithms and recommendation approaches based on psychological theories. The field has its own yearly RecSys conference that also hosts a substantial amount of industry work on this topic.

Recommender LAB at JADS

The recommender LAB at JADS is led by Martijn Willemsen, associate professor at TU/e and JADS. The LAB approaches recommender systems from an interdisciplinary perspective, with a strong focus on the application of decision psychology on algorithmic development and on user evaluation. The LAB is part of the Data and Humans research program that focuses on topic such as causal data science, algorithmic transparency and understandable models.

The recommender LAB host researchers at TU/e and JADS working on different applications of recommender systems in both classical entertainment domains such as movie and music recommendations, as wel as in domains related to behavioral change such as energy conservation and e-coaching and life-style recommendations. Check for details the different topics on the right.

User-centric Evaluation Framework

The recommender LAB recognized that proper evaluation of recommender systems goes beyond offline testing of algorithmic accuracy or behavioral AB tests. The research in our lab typically employes the user-centric evaluation framework developed by Bart Knijnenburg and Martijn Willemsen, in which subjective system aspects (user perceptions) and evaluations (e.g. choice and system satisfaction) are related to objective behavioral measures of user interactions with the system when specific versions of a system are tested against eachother. This approach allows for a more thorough understanding of why and how a particular recommendation approach works (or not).

User-centric Evaluation Framework

Relevant research papers:

  • Knijnenburg, B. P., & Willemsen, M. C. (2015). Evaluating Recommender Systems with User Experiments. In F. Ricci, L. Rokach, & B. Shapira (Eds.), Recommender Systems Handbook (pp. 309–352). Springer to springer
  • Knijnenburg, B.P., Willemsen, M.C., Gantner, Z., Soncu, H., Newell, C. (2012). Explaining the User Experience of Recommender Systems. User Modeling and User-Adapted Interaction (UMUAI), vol 22, p. 441-504, open access