Legal professionals often face challenges in their daily work due to the rapid generation of new legislation and the utilization of unstructured formats unsuitable for automated processing. Consequently, a significant amount of heterogeneous and chaotic information is produced, leading to information overload. To address this issue, we propose a novel model that combines state-of-the-art neural architectures for language processing with classical fuzzy logic techniques. In this study, we evaluate the efficacy of this model using the lawSentence200 benchmark dataset, and the preliminary findings indicate promising results.