Research Center for Digital Sustainability

Negation Scope Resolution for the Multilingual Legal Domain

Negation Scope Resolution for the Multilingual Legal Domain

This project is available as a Seminar or Bachelor's project. This project is available as a group project.


Swiss court decisions are anonymized to protect the privacy of the involved people (parties, victims, etc.). Previous research [1] has shown that it is possible to re-identify companies involved in court decisions by linking the rulings with external data in certain cases. Our project tries to further build an automated system for re-identifying involved people from court rulings. This system can then be used as a test for the anonymization practice of Swiss courts. For more information regarding the overarching research project, please go here.

Negation Scope Resolution describes the task of finding the scope of a negation cue (e.g. "not") in a sentence (see example below). Recent transfer learning techniques such as BERT have been successfully applied on the Negation Scope Resolution task [2]. Negation Scope Resolution has been investigated on multilingual data [3], but, to the best of our knowledge, not yet in the legal domain.

Decision 8C_873/2015 {T 0/2}:

dass der vorliegende Begründungsmangel offensichtlich ist, weshalb auf die Beschwerde in Anwendung von Art. 108 Abs. 1 lit. b BGG nicht einzutreten ist
dass von der Erhebung von Gerichtskosten für das bundesgerichtliche Verfahren umständehalber abzusehen ist (Art. 66 Abs. 1 Satz 2 BGG), 
dass in den Fällen des Art. 108 Abs. 1 BGG das vereinfachte Verfahren zum Zuge kommt und die Abteilungspräsidentin zuständig ist,  

Negation Cue underlined, Negation Scope in italics

Research Questions

So far, to the best of our knowledge, the performance of Negation Scope Resolution is still subpar in multilingual legal documents

RQ1: What is the performance of current negation scope resolution systems on multilingual legal data

RQ2: To what extent can we improve upon this performance?


  1. Get roughly familiar with the literature on Negation Scope Resolution
  2. Select suitable data (diverse legal data (contracts, court decisions, legislation, commentaries, etc.) in multiple languages)
  3. Setup prodigy for negation scope resolution annotations and write annotation guidelines
  4. Annotate and/or supervise the annotation process
  5. Evaluate the performance (probably token-level F1) of existing systems on the newly annotated dataset
  6. Train a new system (probably based on transformers) and evaluate it
  7. Analyze the results


⬤⬤⬤◯◯ Programming

⬤⬤⬤⬤◯ Experimentation

⬤◯◯◯◯ Literature


Good programming skills (preferably in Python)

Preferably experience in deep learning (transformers)


Joel Niklaus


[1] Vokinger, K.N., Mühlematter, U.J., 2019. Re-Identifikation von Gerichtsurteilen durch «Linkage» von Daten(banken). Jusletter 27.
[2] Aditya Khandelwal and Suraj Sawant. 2020. NegBERT: A Transfer Learning Approach for Negation Detection and Scope Resolution. In Proceedings of the 12th Language Resources and Evaluation Conference, pages 5739–5748, Marseille, France. European Language Resources Association.
[3] Anastassia Shaitarova and Fabio Rinaldi. 2021. Negation typology and general representation models for cross-lingual zero-shot negation scope resolution in Russian, French, and Spanish.. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop, pages 15–23, Online. Association for Computational Linguistics.