Daniel Braun

@d4br4
117 Followers
134 Following
76 Posts
Professor of Computer Science at the University of Marburg. Working on Natural Language Processing (NLP), AI for Social Good, and Legal Tech.
Research Group Websitehttps://www.responsible-nlp.net
FotoMartin Schäfer
Alles Gute zum #GIburtstag! Als GI-Mitglieder gestalten wir gemeinsam die digitale Zukunft. 💛
Lust mitzumachen? Hier gibt's alle Infos zur Mitgliedschaft: https://gi.de/mitgliedschaft #WirSindInformatik
This week, I’m in Heidelberg attending the Heidelberg Laureat Forum #hlf25. It’s always a great pleasure to be around some of the smartest minds in Math and Computer Science and see what they are working on.
Today I had the pleasure to talk about label variation in legal data annotation at Ruhr Uni Bochum as part of their AI & Law Seminar Series. Very interesting discussions and RUB is definitely the place to be if you like brutalist architecture.
Verwaltungsanalogisierung

I’m in Bangkok for #ACL2024 and will talk about Legal NLP:

On Monday I present our corpus and paper AGB-DE (https://arxiv.org/abs/2406.06809) at the third poster session.

On Thursday I will talk about teaching NLP in Law School as part of the Workshop on Teaching NLP.

If you’re at ACL and interested in Legal NLP (or NLP 4 Social Good, Teaching NLP, Disagreement and Perspectivism, other cool applications or just want to have a chat) stop by!

AGB-DE: A Corpus for the Automated Legal Assessment of Clauses in German Consumer Contracts

Legal tasks and datasets are often used as benchmarks for the capabilities of language models. However, openly available annotated datasets are rare. In this paper, we introduce AGB-DE, a corpus of 3,764 clauses from German consumer contracts that have been annotated and legally assessed by legal experts. Together with the data, we present a first baseline for the task of detecting potentially void clauses, comparing the performance of an SVM baseline with three fine-tuned open language models and the performance of GPT-3.5. Our results show the challenging nature of the task, with no approach exceeding an F1-score of 0.54. While the fine-tuned models often performed better with regard to precision, GPT-3.5 outperformed the other approaches with regard to recall. An analysis of the errors indicates that one of the main challenges could be the correct interpretation of complex clauses, rather than the decision boundaries of what is permissible and what is not.

arXiv.org
I’m at #JURIX in Maastricht this week, starting with the Workshop on Legal Data Annotation.
Mit dem #CGFx2023 der @informatik war ich in den letzten beiden Tagen auf dem #DigitalGipfel um für unsere gemeinsamen Positionen zum Thema Digitalisierung einzutreten. Es waren zwei intensive Tage und überall war das Fazit es ist schon viel geschafft aber auch noch viel zu tun. Gerade im Bereich der Verwaltung sind wir immer noch weit von dem entfernt was der Anspruch sein sollte und was auch Gesetze vorgeben. Irgendwann müssen wir aufhören zu diskutieren und anfangen zu machen.

In der Politik ist man bekanntlich bis 35 jung. Deshalb durfte ich heute beim Common Grounds Forum der @informatik mit anderen jungen Menschen unsere Ideen zur Digitalpolitik diskutieren.

Besonders gefreut hat mich, dass mich die Teilnehmer:innen als einen ihrer Vertreter gewählt haben um die gemeinsam erarbeiteten Positionen im November beim Digital-Gipfel der Bundesregierung in Jena zu vertreten.

Today I’m at Maastricht University for an exciting seminar on the Use of Artificial Intelligence in International Criminal Courts organised by Africa Legal Aid. Interesting to hear the thoughts of ICC judges on the use of AI in legal proceedings.
Something I was looking forward to all week: The panel on Generative AI at #HLF23.