Keith van Rijsbergen Award

Keith van Rijsbergen has been a pioneer in modern information retrieval and a strong advocate of the development of models and theories in information retrieval. Keith has also been recognised for his mentorship of numerous researchers within the community.

ECIR 2024 proposes a new award (the KvR Award) to encourage and recognise researchers who have made significant contributions in using theory to develop new information retrieval models, paradigms, concepts or metrics, thereby addressing foundational aspects in information retrieval and opening up new ways of thinking about information retrieval.


The KvR Award is open to all active researchers in information retrieval, who have had at least 10 years of experience since their PhD.


The awardee must have had a significant track record in information retrieval, especially in using theory to advance our understanding of key aspects, problems or applications in information retrieval.

An inaugural call for nominations was made in December 2023. A panel, including representatives of the University of Glasgow decided on the awardee.

Awardee 2024: Prof. Maarten de Rijke, University of Amsterdam & ICAI

The panel was particularly impressed by Maarten’s sustained efforts over the years in ensuring strong theoretical foundations in his diverse and numerous contributions to information retrieval, as well as his sustained efforts in pushing the boundaries of the field.
Nominator: Avishek Anand

Award Keynote Talk: List-wise Feature Attribution for Explaining Ranking Models

Tuesday 26th, 1630

Feature attribution explanations are a post-hoc family of explainability approaches that assign scores to each feature, indicating their relative contribution to a model decision. In the talk, I define the notion of feature attribution for ranking models, and list essential properties that a valid attribution should have. I then propose RankingSHAP as a concrete instantiation of a list-wise ranking attribution method. I argue that because of the contrastive nature of the ranking task, list-wise feature attribution can be a powerful and flexible tool for gaining insights into model decisions.
Based on joint work with Maria Heuss and Avishek Anand.