How to inquire about perceived transparency in user studies on recommender systems (Bachelor or Master Komedia)

[This Bachelor/Master thesis can be written either in German or English.]

Nowadays, recommender systems (RS) are a pervasive presence in everyday life: many e-commerce, streaming services and social media apps facilitate users’ decision making by using recommender algorithms. Despite important achievements in the field, most RS are still perceived as black boxes by users, where little can be done to obtain the reasons that justify the recommendations. However, the perception of transparency of a recommender system can be increased by means of providing explanations.

Previous work has addressed the effect of different types of explanations in review-based RS, on the perception of different explanatory aims (e.g. transparency, effectiveness, trust) by users. While useful items from various questionnaires have been used for the evaluation of effectiveness and trust, we believe a more adequate set of items to address the perception of transparency by users is still needed. In this respect, the question of what can be a reasonable indicator of transparency of a RS remains open. For example, Knijnenburg et al. 2012 defines their transparency construct in terms of the user’s understanding of why something was recommended. However, we believed that the perception of transparency of a RS by users could be inquired in users studies in a more thorough and adequate manner. In consequence, the main aim of this bachelor/master work is to develop and test a proposal of a questionnaire to assess the transparency of explanatory RS.

If you are interested in this work, please contact Diana C. Hernandez Bocanegra (enclosing a current overview of grades).



Related focus areas