Pattern-based parsing of German traffic regulations (StVO) for legal knowledge graph construction

Authors: Ibrahim Siddig, Sviatoslav Tugeev, Munir Georges

Abstract:

: In this work, we present a domain-specific pattern-based parser (Π) designed to extract entities and relationships from the German traffic regulations legal text (StVO) using explicit citations and hierarchies within the text to construct a legal knowledge graph (KG). The domain knowledge embedded within the constructed knowledge graph (KG) is a valuable asset for a Large Language Model (LLM) training as part of future research in this domain. To the best of our knowledge, there has been no specific focus in the literature on parsing and tagging legal entities and their relationships for constructing knowledge graphs (KGs) from the German traffic regulations legal text, Straßenverkehrs-Ordnung (StVO), while focusing specifically on the hierarchical structure of the document. While other methods of extracting legal and factual knowledge from similar documents have been attempted, they either focused on different methodologies or different domains. The results indicate that the parser is capable of constructing a highly interconnected graph while maintaining the integrity of the original StraßenverkehrsOrdnung (StVO) legal text, despite the graph’s complexity.