Researchers have developed CLASE, a novel hybrid method for evaluating the stylistic quality of Chinese legal text generated by large language models. This approach combines linguistic feature-based scores with LLM-as-a-judge scores, learning from contrastive pairs of authentic legal documents and their LLM-restored counterparts. CLASE aims to capture both surface-level features and implicit stylistic norms in a transparent, reference-free manner, demonstrating higher alignment with human judgments than traditional metrics and pure LLM-as-a-judge methods. The system also provides interpretable score breakdowns and suggestions for improvement, offering a practical solution for professional stylistic evaluation in legal text generation. AI
IMPACT Provides a more accurate and interpretable method for evaluating AI-generated legal text, potentially improving the quality and trustworthiness of LLMs in legal applications.
RANK_REASON Academic paper detailing a new method for evaluating AI-generated legal text. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- Chinese LegAlese Stylistic Evaluation
- CLASE
- DagsHub
- Gotit.pub
- Hugging Face
- ScienceCast
- Yiran Rex Ma
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
