نوع مقاله : مقاله پژوهشی
موضوعات
عنوان مقاله English
نویسندگان English
Field and Aims: The growing prison population, the heavy costs of inmate maintenance, and the ineffectiveness of imprisonment‑oriented policies have underscored the necessity of revising Iran’s criminal justice model and adopting a rational, data‑driven use of non‑custodial and rehabilitative measures. In this context, advances in artificial intelligence (AI) and machine learning offer new opportunities for more precise decision‑making, risk prediction, and individualized judicial responses. Accordingly, the present study aims to design a localized model of an “Intelligent Judicial Assistant for Alternatives to Imprisonment” that utilizes data analytics and explainable algorithms to assess the risk of recidivism and recommend the most appropriate non‑custodial action for each offender.
Method: This research employs a descriptive–analytical and comparative approach between prominent international risk‑assessment systems (such as COMPAS and HART) and Iran’s legal framework. Based on the principles of sentence individualization and restorative justice, a three‑layer architecture was developed, including: (1) a data input layer (criminal, social, and economic variables), (2) an analytic layer using explainable AI algorithms (SHAP and LIME), and (3) a legal–cultural output layer aligned with Islamic principles of justice. Simulated datasets were used for the model’s initial testing and accuracy evaluation.
Findings and Conclusions: The findings indicate that the proposed model can estimate the risk‑of‑recidivism score with over 80 % accuracy by weighting individual, social, and economic variables. The integration of a legal–cultural filter ensured full compliance of the outcomes with Islamic judicial principles and the judge’s discretion. Moreover, the implementation of a dynamic feedback loop enhanced the model’s continuous learning and precision. The intelligent judicial assistant could, in practice, reduce prison overcrowding, enhance consistency in the application of non‑custodial measures, and foster data‑driven justice grounded in human dignity. Full realization of this potential requires the establishment of clear ethical and regulatory frameworks, regular algorithmic audits, and an independent supervisory body to monitor discrimination and bias in accordance with Islamic principles of justice and transparency.
کلیدواژهها English
