Predicting the outcome in sentencing in Danish criminal cases
When a person is charged in a crime, it is important to have easy access to reliable information about the sentencing to expect. ‘Ilas Vejrudsigt’ (‘weather forecast’) is founded on data and machine learning (ML). A model has been developed to predict the expected outcome in criminal cases. It can be reassuring knowledge in a situation full of uncertainty.
Founder of ilas Aps and the idea behind ‘ilas Vejrudsigt’is Rasmus Wandall. He has a PhD., is lawyer (L) at the law firm Jon Palle Buhl and senior research fellow at Lund University, and he has more than 20 years of experience with criminal law and criminal process in Denmark and internationally. His deep commitment and understanding of people's need for concrete and reliable knowledge, when dealing with the legal system, is the cornerstone of ilas ApS.
Every year in Denmark, more than 300,000 people and relatives go through a criminal process involving the police, lawyers, prosecutors and sometimes the court. For each person, it is a big burden, and it is quite natural to lose track of what can and will happen. The consequence is that many come to a standstill with their lives; withdraw from work, friends and family, and stop planning things for the future. The challenge is, you have no chance of knowing what to sentencing to expect, when you are charged with a crime.
“It is one thing to be told what the applicable penalty framework is for similar cases. To have knowledge of what is common practice is something else. Just because the penalty says up to 6 years in prison, does not necessarily mean that it is likely that you will go to prison," explains Rasmus Wandall, CEO, Ilas ApS.
Before the accused or the relative turns to a lawyer, a first knowledge of what to expect can be reassuring. It gives the individual the opportunity to deal with their specific situation and the potential consequence, it could have for their family and their job. Naturally, this requires that you have confidence in the information and maintain your anonymity
Rasmus Wandall, CEO, Ilas ApS.
To know the applicable penalty framework for similar crime case is one thing, but having knowledge of what is common practice is something else.
When the person entering answers to 11 questions about their situation, the probability of whether the person will be acquitted or found guilty is predicted. The person can then choose to continue to get insight on the expected sentencing being a fine, community service, conditional or unconditional imprisonment. If it is a matter of prison, the expected length will also be indicated.
"The solution uses the person’s entered information completely anonymously. Answers entered are not saved, nor does ilas.io register, who uses the solution” explains Rasmus Wandall.
The model of the prediction, which is the foundation of ’Ilas Vejrudsigt’, is product developed based on research by Christian Møller Dahl and Simon Friis Wittrock at SDU in collaboration with Lund University. The model was trained and tested on the basis of data from Statistics Denmark, with all criminal decisions in Denmark in the five-year period from 2013-2018 included.
“In the development of the model, it is absolutely central that the result does not become a 'Black Box'. The results must be understandable and validated to ensure, that they are representative and trustworthy. Therefore, a significant part of the development of the model has been to validate and interpret the results, among other things by using Explainable AI," says Christian Møller Dahl.
With ‘ilas Vejrudsigt’, accused persons and their relatives get a better tool to obtain an insight into how cases usually end. Of cause. the answers cannot stand alone, and the normative assessments that a lawyer can contribute with can still be relevant. That is why Ilas.io also offers the option of using 'Find Lawyer', which makes it possible to search for lawyers, who have experience in the relevant legal area.
Even for lawyers, there may be crime areas, where they do not have in-depth experience, which is time consuming to uncover to be able to give an indication, of what outcome the accused can expect based on previous cases. Therefore, 'ilas Vejrudsigt’ is also an analytical and quantifiable tool for lawyers, which provides a unique opportunity to get concrete figures on the table.
In parallel with the development of the product, the article ‘Predictions of Guilt and Criminal Sentencing’ has been published in Nordisk tidsskrift for kriminalvidenskab – nr. 1/2023 (Nordic Journal of Forensic Science), which explains the results of the model.
”Three parties are involved in the development of the product. My focus is to create a relevant product that helps people in need and contributes to the democratization of the legal process. THINKcreative ApS ensures the development of an understandable and intuitive user interface. Rooftop Analytics uses the research results to implement the prediction model on Ilas.io. They also managed the project, which ensured a smooth process, with high professionalism, transparency and knowledge synergies," tells Rasmus Wandall
The solution is based on Danish data (and law) and developed in Danish only.
ABOUT Rooftop analytics
With Rooftop Analytics' ambition to perform 'knowledge transfer', it was obvious to join the project financial upported from Innobooster, and later to take co-ownership, when Ilas ApS was founded in 2019.