DAS Rechtsbijstand
Automatic encoding of documents
DAS handles legal cases and provides legal expenses insurance. DAS had a clear ambition to bring their historical dataset of 400,000+ legal documents up to date with a new coding system.
The challenge
DAS handles legal cases and provides legal expenses insurance. DAS had a clear ambition to bring their historical dataset of 400,000+ legal documents up to date with a new coding system. They saw that employees were spending a lot of time assigning the new coding to incoming documents, and were looking for a way a predictive algorithm could support them in this. The new coding consisted of 4 different parts and ended up having 1200+ different variants.
What we did
In consultation, we came up with a Natural Language Processing (NLP) project. An algorithm reviews text from legal documents to predict the most likely codes. With the algorithm’s suggestions, both old and new documents can be correctly coded faster, according to the new system. The challenge was divided into 2 projects, with each project developing a model for 2 of the 4 components of the new coding. We developed a predictive model for each of the 4 parts of the new coding, indicating for each document the most likely complete coding including a certainty percentage.
The result
Documentation was delivered explaining the structure of the models, including what choices were made within them and why. Follow-up steps have been defined for continuous improvement of the model as a transfer to the data science team within DAS.
Test your own data maturity
Test your own data maturity
Your data maturity determines the results you can get from your data. How ‘data mature’ are you? By knowing where you stand, you can better assess what is the logical first step. Can’t figure this out by yourself? We are happy to help you discover it via our free online quick scan.
Wondering where your company stands? Take the online quick scan here!
Why collaborate with JADS MKB Datalab?
The JADS MKB Datalab supports Dutch SMEs that want to make the transition to a more data-driven way of working. The JADS MKB Datalab carries out projects with SME entrepreneurs and Data Science students. In addition, they offer several tools to help you define a Data Science project with impact, such as workshops, a Data Maturity Scan and a Data Project Canvas.
So, if you want to take on a specifically designed short-term data science project for your SME with a JADS MSc student, visit the JADS MKB Datalab Website. 150 SMEs have already participated!