top of page

Justice Innovation II | Blog

Regular clients have always been quite ignorant about the specifics of the process and of the legislation of their own case. Recent ongoing research focussing on data-driven innovation has led to the preliminary development of a legal recommender system [1]. In this article, an algorithm has been put forward that offers an easier way to get access to relevant case law, by giving users of a legislative portal, suggestions of

other relevant sources of law and cases, given a focus legislative article. Not only that, but it also adapts the ranking of relevance of other articles (sources of law) related to that case law. This system will allow different type of users to make use of it.

As a judge, having quicker and broader access to other case law, helps making better, less biased decisions, reducing the subjectiveness of a decision to different interpretation or just personal belief by different judges. As a layer, it facilitates the search. Most importantly, as a client, on one hand, the fact that the layer has faster and easier access to the law and to similar cases reduces the amount of time he spends working on that client’s case, hence reducing the price for hiring the layer. On the other hand, if such a system is made available to the public this also allows for a whole new ball game, where the client is able to become more informed about his own case and in a way, keep track of his layer’s work, tackling the big issue of lack of client power. However, making it available to the public could possibly arise another entirely new complex challenge and that is of personally and demographically identifiable information. Case law may contain sensitive and private information about an individual, an organization or a group of people, that in such a case could be seen by another person. There are a lot of risk factors involved with this, which highlights the fact that one needs to use the data responsibly.

There are prospects of further improving this algorithm by also exploit the network of case law itself. This could be used to estimate the authority of cases, hence improving the suggestion of relevant case law.

However, this is undoubtedly a very sensitive data-driven innovation, which should be further studied in a responsible manner and the data involved in it should also be used in a responsible way. The research presented in [1] only focused on creating such a network between the legislation and the case law for a specific field of law. In the article it is stated that the date of the decision and the court are also relevant for the purpose of this innovation which leads one to think that that may be the next step to take in the near future. Caution needs to be taken here. If one continues into expanding the concept into including date of decision, this may be problematic, as there are times where lots of cases within a field of law are occurring at the same time and this may fog and influence the judgement of a judge at that particular moment in time. A prime example of this is happening right now with all the asylum seekers that are fleeing to Europe due to conflicts in the middle east and especially in Syria.

Furthermore, one should not forget that ultimately the algorithm mentioned is a mere piece of code. Even though the advantages of this new legal recommender system are clear, one should not discard the possibility that errors might occur in the code and in the resolving process, leading to an unfair trial and decision by a judge. In other words, a false correlation may lead to a negative causation. In fact, in the report, the samples tested did have ‘false’ positive, even though they were between 1-5% of the total sample space, they should not be neglected.

[1] Winkels, R., Boer, A., Vredebregt, B., & van SOMEREN, A. (2014, November). Towards a Legal Recommender System. In JURIX (pp. 169-178).

bottom of page