Opening the Black Box: Increasing Intelligibility of Recommendations via Explainable AI Techniques (Master Komedia/Informatik)


Recommender systems (RS) have recently experienced substantial progress in terms of algorithmic sophistication and quality of recommendations. Yet, RS mainly act as black boxes and are limited in their capability to explain why certain items are recommended. Usually, however, RS have access to various information sources such as images or user reviews that allow for the generation of interpretable recommendations. The goal of this thesis will be to explore and investigate the applicability of such sources to techniques related to explainable AI [1] in order to increase intelligibility of recommendations.

[1] Gunning, D. (2017). Explainable artificial intelligence (xai). Defense Advanced Research Projects Agency (DARPA), nd Web.

(The thesis can either be written in German or English.)


Tim Donkers