Human-in-the-Loop: Legal Knowledge Formalization in Attempto Controlled English

Automating legal knowledge through the use of computational languages presents a series of challenges, particularly in translating complex normative texts into formalized representations that are semantically faithful and computationally valid. A salient concern is reconciling the ambiguity inherent in natural legal language with the syntactic constraints of controlled languages. This work proposes a human-in-the-loop methodology for formalizing normative texts in the controlled language Attempto Controlled English (ACE), centered on the interaction between legal experts and computational constraints. Specifically, key legal provisions and fundamental case-law principles were translated into ACE. The findings reveal that even ostensibly straightforward legal provisions necessitate a substantial decomposition, abstraction, and adaptation process. The analysis also identified recurrent linguistic patterns (modal, temporal, and conditional) and reusable translation strategies.