Chapter Valutazione della propensione alla mediazione tramite eXplainable AI

Abstract

Mediation in civil trials can effectively resolve disputes outside of court proceedings, easing the burden on the courts if successful. Efficiency in identifying disputes is essential, as a failed attempt at mediation can lengthen the duration of the trial. The decision rests with the judge/tribunal on the basis of numerous documents that contain certain statements significant to the decision. This paper describes an artificial intelligence, AI, solution to provide a decision support system that can process documents and (i) produce reliable suggestions, (ii) produce substantiated reasons by highlighting the statements that led to the suggestion, and (iii) respect privacy and data security. Explainable AI techniques (XAI) technologies were used for this purpose, resulting in a solution that meets the defined objectives. The solution was developed as part of the research project "Agile Justice," funded in the Italian National Governance and Institutional Capacity NOP, and validated against real cases. The solution leveraged the Snap4City framework for data management and AI/XAI solution.