Advancing Sustainable Justice through AI-Based Case Law Analysis

Abstract

Background: Ukraine has a unique Unified State Register of Court Decisions that publishes all court decisions in cases considered and resolved by courts in the public domain. There are more than one hundred million such documents in the register today. This provides unique opportunities for collecting, analysing, and summarising the empirical base of justice. This has the potential to form the basis for further transformation of the national model of justice. This study's impetus may have risen from the realisation that relying solely on human resources for such endeavours may present challenges. Methods: The study is based on the hypothesis that using hardware and software to analyse large data sets of state registers of court decisions and judicial statistics data can identify persistent patterns and causes of inefficient functioning of the judicial system. Results and Conclusions: The study led to the development of software with functionality that annotates court decision text, intended for further use in advanced Natural Language Processing algorithms. Furthermore, the study underscores the need to develop an algorithm for predicting risks and outcomes of court proceedings and a methodology for processing large amounts of data from the Unified State Register of Court Decisions. This is justified based on specific indicators of the effectiveness of dispute resolution. This article advocates for the use of machine learning algorithms as an innovative tool to generalise large data sets from court decision registers, particularly to obtain objective data on a large scale. The article also examines the prerequisites for establishing the Institute of National Judicial Practice and explores its functioning in the present stage of judicial reform.