Інформаційно-аналітичні та алгоритмічні інструменти оцінки ризиків: порівняльна характеристика

Abstract

У статті досліджуються сучасні підходи до оцінки ризиків у сфері пробації. Проаналізовано міжнародні та національні моделі оцінки ризиків, зокрема алгоритмічну систему Correctional Offender Management Profiling for Alternative Sanctions і концептуальну модель Multi-Task Dual-Theory Probation Prediction Model, визначено їх можливості та обмеження. The article examines contemporary approaches to risk assessment in the field of probation within the context of the digital transformation of criminal justice systems. Special attention is paid to information-analytical and algorithmic tools used to support decision-making at the stages of pre-sentence reporting, supervision, and correctional planning. The study provides a comparative analysis of international and national risk assessment models, including the algorithmic system COMPAS (USA), the Ukrainian information subsystem "Kassandra", and the conceptual Multi-Task Dual-Theory Probation Prediction Model (MT-DT). The research identifies fundamental differences between actuarial-algorithmic models typical of common law jurisdictions and legally oriented analytical models compatible with continental legal systems. It is substantiated that systems such as COMPAS, which rely on opaque mathematical algorithms and numerical risk indices, raise significant concerns regarding transparency, explainability, procedural fairness, and compliance with the principles of legality and individualization of punishment. These limitations are illustrated through an analysis of the landmark case Loomis v. Wisconsin, which highlighted the risks associated with the use of non-transparent algorithmic assessments in judicial decision-making. The article argues that the Ukrainian probation system currently relies on information-analytical tools that primarily perform accounting and documentation functions and lack autonomous predictive capabilities. In response to this limitation, the author substantiates the feasibility of implementing an integrated risk assessment model based on the MT-DT concept, which combines formalized legal logic with adaptive predictive algorithms.