Poster Presentation at NetSci 2025 - ParaGraph: Analyzing legal citation networks across document types

Date:

At NetSci 2025, I presented the current state of my research project ParaGraph. In this project, we aim to find out what insights can be gained about statutes and decisions from studying how paragraphs of decisions reference sections of statutes.

See the poster at this link!

Abstract

Legal systems are built on text but act through people. In different branches of government (i.e., legislature, executive, and judiciary), individuals or institutions create and interpret legal documents, continuously producing new outputs that feed back into the system and shape the law itself. This process might be bound to formal rules, but the substantive changes to the law do not follow predetermined patterns. Instead, they emerge from the complex interplay between texts, individuals, and institutions. To gain a quantitative understanding of complex legal systems, researchers have used the perspective and methods of network science. Most of their studies, however, have investigated relations within documents of the same type, such as statutes or court decisions, leaving the interplay between the different branches of government out of sight. With ParaGraph, we develop methods for cross-document-type analysis of legal document networks. Our initial dataset includes 61,708 court decisions, published between 2010 and 2024 by Germany’s highest federal courts, as well as all German federal statutes and regulations in their current consolidated form. In our baseline analysis, we find that established (e.g., centrality) measures are not sufficient to capture the nuances in our legal network data. Against this background, we propose two approaches to enhance the legal-network-analysis toolkit: (1) leveraging structural metadata to compute expressive metrics, and (2) confining the analysis to a relevant scope, depending on the domain question of interest. We test our model and methods on their ability to validate a set of legal assumptions, taken from normative sources, and aim to provide a flexible framework that enables novel insights into legal systems.