Masters's Thesis Jieyi Zhang

Title: Automatic Norm Chain Generation for German Legal Verdicts

Background

In this work, we try to automatically generate a legal norm chain for each verdict documentation. A legal norm consists of law code abbreviation, article number, paragraph number, etc. Those legal norms might have different granularity. A norm chain for a document includes the most relevant legal norms for the case and it represents a link between legal norms that explicitly or implicitly reference each other.

Traditionally, a legal norm chain for a verdict document is generated manually by the experts based on their domain knowledge. Therefore, in our work, we try to automate this manual process with a rule-based algorithm as well as machine learning or deep learning models.

Goal

• Research about the state-of-the-art NLP techniques for keyphrase extraction
• Design and implement a rule-based approach as well as ML or DL approaches to extract and assign the legal norm chain for the verdict documentation.

Research Questions

• How are the norm chains created by judges/legal authors?
• How to technically generate the norm chains for each verdict document?
• Can norm chains be generated just by the context in the respective verdict document? Do we need additional input information?