Mapping Sentencing Research: An AI-Assisted Journey to a Comprehensive Overview

• In Event: Re-thinking sentencing research: new possibilities and new approaches

Abstract

Sentencing research spans multiple disciplines, leading to fragmentation and conceptual inconsistencies. To address this, we are conducting a comprehensive scoping literature review as part of the ERC Starting Grant project Sentrix, using a systematic approach that provides an automated (qualitative) evaluation of the sources and their visualization in a network. This paper presents our methodology.
We first developed a categorization framework for the research literature, including properties like year, country of origin, employed research methodology, and research themes and subthemes. AI tools were leveraged to automatically categorize the analysed sources. To ensure reliability and contextual accuracy, we have implemented double-check safety mechanisms, including verification by native speakers and random checks of AI-analysed papers. The sources were then visualized in a network, with the connections between them based on their (dis)similarity.
Our initial focus is on Europe, but the scoping review is designed for potential expansion. We aim to include as much scholarship as possible and draw connections between different strands of research through a detailed weighting of parameters, including fields, themes, subthemes, and keywords. This enables us to track the evolution of sentencing scholarship and identify dominant frameworks.
Our approach balances traditional qualitative review methods with computational techniques to enhance efficiency and scalability. By leveraging AI, we can overcome language and quantity barriers, allowing for a broader and more inclusive analysis of sentencing research worldwide. Furthermore, we examine the role of AI in addressing limitations in traditional literature reviews, enabling analysis of a higher number of articles (in different languages), while acknowledging the need for human oversight to mitigate potential errors in AI-assisted classification.
This approach provides sentencing researchers with an analytical framework for systematically engaging with existing scholarship. By mapping key trends, debates, and methodological developments, we aim to inform future research directions and promote interdisciplinary dialogue in sentencing studies.

Authors

• Mojca M. Plesničar, Institute of Criminology at the Faculty of Law Ljubljana

• Marko Balažic, Institute of Criminology at the Faculty of Law Ljubljana / Inštitut za kriminologijo pri Pravni fakulteti v Ljubljani

• Manja Skočir, Institute of Criminology at the Faculty of Law Ljubljana

• Miha Marčenko, Institute of Criminology at the Faculty of Law Ljubljana

• Dean Lipovac, Institute of Criminology at the Faculty of Law Ljubljana