Hybrid AI for Supporting the European Drafting Legislation

Michele Corazza, Salvatore Sapienza, Monica Palmirani, Generoso Longo, Leonardo Zilli, Emanuele Di Sante

Abstract: This study investigates whether AI and ML tools can support European legislative drafting with advanced information retrieval using AI and which methods are more effective. The article shows the hybrid methods leveraging the combination of LegalXML-annotated (Akoma Ntoso) documents and Natural Language Processing (NLP) techniques that take advantage of the structure of the legislative text to perform their tasks. In particular, the experiments proposed in this paper deal with three crucial functionalities for retrieving relevant legislative information with incomplete inputs using thematic similarity: normative references, legislative definitions, and legislative argument/topic search. The study shows that computational approaches combining LegalXML-based documents and NLP techniques can fruitfully be deployed to support legal drafting tasks in legislative European institutions.