arXiv subjects

Simon Alexander Nonn

Publications and source records attributed to Simon Alexander Nonn.

4 recordsLinked to original sources

arXiv2026

GradeLegal: Automated Grading for German Legal Cases

Grading German legal exam solutions faces growing volumes and a shortage of qualified graders, delaying feedback and creating a bottleneck. At the same time, it is a high-stakes expert task, since state exam grades strongly influence career outcomes in Germany. Despite this practical relevance, literature lacks systematic studies on effective methods for grading legal exams. To address this gap, we investigate whether large language models (LLMs) can support the automated grading of German legal case solutions in criminal and public law, thereby enabling scalable feedback and student self-testing. We present a systematic evaluation of 27 proprietary and open-source LLMs, benchmarking prompting strategies that incrementally add task-related information, such as a sample solution and a grading rubric. Using quadratic weighted kappa (QWK), reasoning-oriented LLMs can approximate expert grading in public law when given a sample solution and a grading rubric (up to 0.91), compared to 0.60 in criminal law, suggesting a harder grading task in criminal law. Beyond single-model grading, ensembling improves agreement by up to 0.15 over its best member and can offer an alternative to stronger closed-source single models. In addition, our findings suggest that effective prompt design and model selection are necessary for reliable LLM-based grading of legal exams.

cs.CL↗

arXiv2026

Robust Generalizable Heterogeneous Legal Link Prediction

Recent work has applied link prediction to large heterogeneous legal citation networks \new{with rich meta-features}. We find that this approach can be improved by including edge dropout and feature concatenation for the learning of more robust representations, which reduces error rates by up to 45%. We also propose an approach based on multilingual node features with an improved asymmetric decoder for compatibility, which allows us to generalize and extend the prediction to more, geographically and linguistically disjoint, data from New Zealand. Our adaptations also improve inductive transferability between these disjoint legal systems.

cs.LG↗

arXiv2025

The Missing Link: Joint Legal Citation Prediction using Heterogeneous Graph Enrichment

Legal systems heavily rely on cross-citations of legal norms as well as previous court decisions. Practitioners, novices and legal AI systems need access to these relevant data to inform appraisals and judgments. We propose a Graph-Neural-Network (GNN) link prediction model that can identify Case-Law and Case-Case citations with high proficiency through fusion of semantic and topological information. We introduce adapted relational graph convolutions operating on an extended and enriched version of the original citation graph that allow the topological integration of semantic meta-information. This further improves prediction by 3.1 points of average precision and by 8.5 points in data sparsity as well as showing robust performance over time and in challenging fully inductive prediction. Jointly learning and predicting case and norm citations achieves a large synergistic effect that improves case citation prediction by up to 4.7 points, at almost doubled efficiency.

cs.SI↗

arXiv2024

On the Suitability of pre-trained foundational LLMs for Analysis in German Legal Education

We show that current open-source foundational LLMs possess instruction capability and German legal background knowledge that is sufficient for some legal analysis in an educational context. However, model capability breaks down in very specific tasks, such as the classification of "Gutachtenstil" appraisal style components, or with complex contexts, such as complete legal opinions. Even with extended context and effective prompting strategies, they cannot match the Bag-of-Words baseline. To combat this, we introduce a Retrieval Augmented Generation based prompt example selection method that substantially improves predictions in high data availability scenarios. We further evaluate the performance of pre-trained LLMs on two standard tasks for argument mining and automated essay scoring and find it to be more adequate. Throughout, pre-trained LLMs improve upon the baseline in scenarios with little or no labeled data with Chain-of-Thought prompting further helping in the zero-shot case.

cs.CL↗