Master thesis student

Email:  yasaman.roshanavandi@stud.uni-due.de

Thesis topic:  Implementation and Evaluation of Different Intelligibility Types in a Visually Explainable Recommender System

Related project: RIMA

Supervisor(s):  M. Sc. Mouadh Guesmi, M. Sc. Shoeb Joarder

Thesis duration:  05 / 2021 - 11 / 2021

Description

Intelligent systems gather inputs implicitly and make decisions based on complex probabilistic models which usually are not understandable to users. This lack of understanding might reduce the transparency of these systems, user trust, and satisfaction. Hence, it is crucial for such systems to be intelligible.

The aim of this thesis is to implement an intelligible system as part of the RIMA project, which allows this system to be capable of explaining its behavior through several interaction types with the end-user. Moreover, investigate and evaluate the impact of different intelligibility types on user understanding and trust under different scenarios while using the RIMA application.