Nationaal Politielab AI – Utrecht


Scientific director

Prof. Dr. Floris Bex

PhD students

Michiel Bron, MSc
Many important facts in police complaints or incident reports are recorded in unstructured free text. This data is difficult to analyse for a computer. These documents must be provided with tags and other metadata to aid analysis. However, this is a time-consuming process if this has to be done manually. My research project aims to create a system that automatically provides these documents with the correct tags. Through the use of new machine learning techniques, we aim to minimise the number of manually tagged documents data needed for the model's training.
Isabelle Fest
My research focuses on the relationship between AI use, citizen trust and public values. How are AI-systems implemented in the organisation and used in practice. For example, how do police officers interact with results or directions given to them by a computer instead of a person? In practice this can differ greatly from what managers and programmers expected in advance, as users interact with the technology. Within this scope, my research focuses on public and ethical values playing a role in the decisions the police employee makes as well as trust citizens have in the police organisation.
Elize Herrewijnen, MSc
When machine learning models are used in practice, e.g. at the Netherlands Police, the explanation and motivation for (the outputs) of the models become relevant. My research focusses on explaining text classification models. In particular on explaining models in line with the algorithmic decision process of the model (faithful), and in a manner that is understandable to users. These explanations come in the form of annotator rationales: (sub)sentences or words from the input text that explain a model’s prediction.
Mijke van den Hurk, MSc
I do research on the field of Social Simulation, where the aim is to build a multi-agent based model that can be used to simulate complex (social) systems. Agents represent humans with needs, identities and relations. These kind of models can give insights on social and dynamical processes, and support policy making within the system. The aim of my research is to build a virtual world that can be used by the Dutch Police for analysis on social-cultural issues or testing of interventions on criminal organizations.
Laurens Müter, MSc
To extract information from social media, I try to find textual and divergence patterns in micro-blogs related to incidents.
Esther Nieuwenhuizen, MSc
I am a PhD candidate at the Utrecht University School of Governance. I have a background in Public Administration, with a MSc in Public Management. In my PhD research, I will look into the effects of transparency about (the use of) algorithmic tools on citizen trust in the Dutch National Police.
Daphne Odekerken, MSc
A dialogue system, or conversational agent, is an agent that communicates with another agent. Dialogue has been studied in the domains of human-computer interaction and formal argumentation. In my dissertation research, I explore the possibilities of combining these two approaches into hybrid argumentative-conversational agents in the legal domain. By combining the strengths of machine learning based conversational agents and formal argumentation, I aim to develop agents that can accurately and efficiently respond to natural language input by asking relevant questions and drawing explainable conclusions. In my function as AI Scientist at the Dutch National Police, I am currently implementing a dialogue system for the automatic intake of complaints of online trade fraud.
Joeri Peters, MSc
My research is related to the idea of intelligent monitoring. To be able to anticipate to developments in society in an early stage, it is important to reason with scenarios and data. For this purpose I apply scenario analysis, argumentation and strategy definition. I am developing formalisms and methods that enable reasoning with scenarios comparable to their application in murder cases: with evidence, alternative stories, alibis etc.
Marcel Robeer, MSc
New techniques in the field of Machine Learning (ML) offer the Dutch Police many opportunities for data-driven analysis and decision-making. However, the complexity of these methods makes it unclear how the decisions made using these ML models were generated from the data. For the Police, it is of vital importance that decisions made are grounded in findings and supported by facts that strengthen the case. In my PhD, I investigate and develop methods for (interactively) explaining decisions made by artificial intelligence (AI) systems, in a human-understandable, useful and truthful manner.
Carlos Soares, MSc
My research focuses on the relationship between algorithms and work. Technological systems can impact the execution of a task. Does an AI system replace an employer or does the system enhance employers’ efficiency? The goal of my research is giving insight in the use of AI systems by the employees.
Martijn van Vliet
Among others, the National Police Lab AI produces new and innovative initiatives for the Dutch National Police. For my PhD I will conduct research to develop a method that can guide these new initiatives through further development phases, with the ultimate goal to deploy them as reliable and available applications that provide added value for the Police. The main focus in this research will most likely be an initiative that uses speech-to-text and other artificial intelligence techniques in order to reduce the administration burden on employees within the organization.

Senior researchers

Dr. AnneMarie Borg
In several of the components of the police project we work with formal argumentation. For this purpose we work with structured argumentation in dynamic settings, something which is relatively unknown. In my research I aim at developing argumentation frameworks that can be implemented in the police project.
Dr. Merijn Bruijnes
With a background in psychology and computer science, I work within the ALGOPOL project on an organizational reflection tool that allows the police to design and evaluate algorithms. The focus is on the transparency and value sensitivity of an algorithm and its impact on the various stakeholders, for example employees and the organization on the one hand, and citizens and society on the other.
Dr. Jelte Mense
My main research at the Police Lab AI revolves around how the police can use AI and data science to become more effective. I am also interested in how the police can leverage open data such as news to detect and anticipate important developments that could be relevant to the Netherlands. Next to data science methods I am also passionate about agent-based modelling and social simulation, for example through exploring the effectiveness of police interventions in criminal networks in computer models.
Dr. Bas Testerink
Teamlead TWO (Driebergen, Odijk, Utrecht)
Team scientific developments (Dutch abbreviation: TWO) is a scientific R&D team at the police which acts as a bridge between academic theory and police practice. We are particularly interested in research regarding A) data acquisition, such as active learning, B) analysis, monitoring and control of systems, such as game theory and normative systems, C) optimalisation, such as reinforcement learning, D) efficient data usage, such as transfer learning.

In addition, I'm also active in the field of A.I. & ethics. I do this from a technological perspective, as I'm not an ethicist myself.

Affiliated researchers

Drs. Dennis Craandijk
Formal argumentation is characterized by some hard reasoning problems which are currently only solvable by symbolic algorithms. However, recent advances in neural networks which can operate on graphs, have opened up the possibility to use learning-based methods on such reasoning problems. My research aims at bridging the gap between symbolic-based and learning-based AI by applying these learning-based methods to logical reasoning problems in argumentation.
Lisa Fijn
My research investigates the behavioural needs of search and detection dogs at the police. Mainly, I study these working animals when they are off duty. As a behavioural biologist, I analyse a combination of behavioural observations and physiological parameters. I use this biometric data to look for the optimum balance between activity and rest in relation to both search performance and animal welfare. Additionally, I look at the housing and management of these dogs and focus on the physical and mental challenges as well as their social needs.
Dr. Maša Galič
As a privacy & criminal procedure law expert, my ambition is to advance the regulation of digital investigation powers of the police, which function at the intersection of criminal procedure law and data protection law. I am also specifically interested in how AI can be employed in the criminal justice domain in a lawful and legitimate way.
Pascalle Roulaux
My research aims to optimize the performance of the scent tracking and detection dogs that are used by the Dutch police. In addition to my education as an animal scientist, my recreational practical experience with scent tracking and detection dogs is valuable for this research. One of the first goals is to objectively determine the qualities of well-performing dogs.

Previous members

Dr. Marijn Schraagen
I am a computational linguist working on a variety of topics in information retrieval and classification. I obtained a bachelor and master degree in Artificial Intelligence from Utrecht University and a PhD in Data Mining from Leiden University. My application areas of interest include computational historical sociolinguistics and NLP in the legal domain.
Remi Wieten, MSc
I obtained my PhD in september 2021 from the Department of Information and Computing Sciences at Utrecht University with a background in Mathematics (BSc) and Forensic Science (MSc). In my project, I am concerned with: (1) developing methods and techniques for the (semi-)automated construction of Bayesian networks (BNs) from arguments; and (2) designing argumentation-based techniques for reasoning about BNs, which can aid experts in validating, refining and choosing between constructed BNs.