Artificial Lawyer recently caught up with Bram Stalknecht, CEO of Dutch AI developer SemLab and founder of JuriBot, a legal AI litigation analysis application capable of criminal trial outcome prediction.
Stalknecht begins by explaining that JuriBot has been built by SemLab, a Dutch company that says of itself: ‘[We are] Europe’s main developers of semantic software applications. Today we have one of the largest groups of experts in Natural Language Processing (NLP), Computational Linguistics and Artificial Intelligence technologies.’
SemLab has around 45 full time staff working on AI. The company operates in several areas and has developed legal applications before, such as CrimiKaart, which allows people to track crimes committed in Holland at a very detailed level.
However, JuriBot, is quite different and seeks to make the life of trial lawyers easier by analysing court cases. The system currently is in Beta and will be offered free to law students later in 2017 and also by subscription to lawyers.
The application includes all public cases for criminal and related case law in the Netherlands. Other case types will be steadily added in the months and years ahead.
At present JuriBot can use NLP to illuminate and structure data found in court documents to:
- Show defendant or perpetrator characteristics.
- Reveal specific sections of case documents.
- Filter cases by judge or lawyer.
- Can show related data on acquittals or convictions.
Juribot adds that by using this data it helps ‘to estimate which lawyer, at which court, for which type of offence, is most likely to achieve an acquittal’. I.e. JuriBot has case prediction capabilities.
Stalknecht says that the inspiration was the challenge to see if they could apply technology they had used in other areas, such as finance, to the law.
‘[The aim was] to turn legal archives into actionable data and insights. We have already successfully developed trading analytics for financial markets based on the same technology,’ he says.
As to how this is different from several other start-ups working in the current red hot litigation analytics space, Stalknecht says: ‘JuriBot is semantic data extraction of relevant case attributes, which we allocate by NLP and machine learning. Then our system compares metadata with the archive and presents this in a prediction analysis dashboard.’
‘Another unique selling proposition is that JuriBot works language independent. Where some systems can only process United States case law, JuriBot is multilingual and is easy to adjust to any country’s specific law,’ he adds.
Although, for now, as noted, the system is focused on Dutch law and criminal case law. And, as there is yet to be a ‘BakeOff’ between the different legal AI litigation analysis and prediction systems on the market it is very hard to say which is better, more user-friendly, or more accurate. But, given the size of the global litigation market it is probably fair to say there is plenty of room for multiple players and that this is still early days.
Also, this Dutch development is further proof that legal AI is not just a North American or UK concern, but is now being explored by pioneers in multiple markets, and in multiple languages. In fact, European legal AI developers, which once did not receive much coverage now seem to be on the rise.