Odysseas S. Chlapanis
PhD student at AUEB · affiliate researcher at Archimedes
I am a PhD student at the Athens University of Economics and Business and an affiliate researcher at the Archimedes Research Unit. My work centers on evaluating and improving how large language models reason over legal text: measuring their legal reasoning and citation faithfulness, and building practical, grounded LLM applications for law. I am also interested in what cognitive science reveals about how these models represent meaning, from legal concepts to abstract words.
Publications
The Grounding Gap: How LLMs Anchor the Meaning of Abstract Concepts Differently from Humans
Preprint, arXiv:2605.08837, 2026
GreekBarBench: A Challenging Benchmark for Free-Text Legal Reasoning and Citations
Findings of the ACL: EMNLP 2025, Suzhou, China
AUEB-Archimedes at RIRAG-2025: Is Obligation concatenation really all you need?
1st Regulatory NLP Workshop (RegNLP 2025), pp. 52-58, Abu Dhabi, UAE
LAR-ECHR: A New Legal Argument Reasoning Task and Dataset for Cases of the European Court of Human Rights
Natural Legal Language Processing Workshop (NLLP 2024), Miami, FL
Local Explanations and Self-Explanations for Assessing Faithfulness in Black-box LLMs
13th EETN Conference on AI (SETN 2024) Workshops
Archimedes-AUEB at SemEval-2024 Task 5: LLM explains Civil Procedure
18th International Workshop on Semantic Evaluation (SemEval-2024), Mexico City
Adapted Multimodal BERT with Layer-wise Fusion for Sentiment Analysis
IEEE ICASSP 2023, Rhodes Island, Greece
Other
Gemini-3 Benchmarkathon Blog
Hugging Face blog, November 2025. Benchmarking Gemini-3 on legal tasks, including GreekBarBench.