Why, vitrivr? Understanding Results in Multimedia Retrieval (Master Project, Ongoing)


Cristina Illi


As Multimedia Retrieval Systems become more advanced and complex and deal with larger amounts of data, the topic of explainability becomes of ever greater importance for two reasons: First, understanding the inner workings of machine learning algorithms often considered blackboxes helps developers to improve the system itself. Second, and often overlooked, by providing information on the retrieval process, users can improve their queries in an interactive and iterative process. In contrast to Relevance Feedback, the objective is helping the user to improve their query instead of improving results based on feedback by the user. This project aims to improve the vitrivr system by adding functionality which helps users understand why each result has been returned.

Start / End Dates

2020/09/21 - 2021/01/31


Research Topics