Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.09990 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908877522468864 |
|---|---|
| author | Begnini, Ana Vicente, Matheus Souza, Leonardo |
| author_facet | Begnini, Ana Vicente, Matheus Souza, Leonardo |
| contents | In business-to-business relations, it is common to establish NonDisclosure Agreements (NDAs). However, these documents exhibit significant variation in format, structure, and writing style, making manual analysis slow and error-prone. We propose an architecture based on LLMs to automate the segmentation and clauses classification within these contracts. We employed two models: LLaMA-3.1-8B-Instruct for NDA segmentation (clause extraction) and a fine-tuned Legal-Roberta-Large for clause classification. In the segmentation task, we achieved a ROUGE F1 of 0.95 +/- 0.0036; for classification, we obtained a weighted F1 of 0.85, demonstrating the feasibility and precision of the approach. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_09990 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | A Two-Stage Architecture for NDA Analysis: LLM-based Segmentation and Transformer-based Clause Classification Begnini, Ana Vicente, Matheus Souza, Leonardo Computation and Language Artificial Intelligence In business-to-business relations, it is common to establish NonDisclosure Agreements (NDAs). However, these documents exhibit significant variation in format, structure, and writing style, making manual analysis slow and error-prone. We propose an architecture based on LLMs to automate the segmentation and clauses classification within these contracts. We employed two models: LLaMA-3.1-8B-Instruct for NDA segmentation (clause extraction) and a fine-tuned Legal-Roberta-Large for clause classification. In the segmentation task, we achieved a ROUGE F1 of 0.95 +/- 0.0036; for classification, we obtained a weighted F1 of 0.85, demonstrating the feasibility and precision of the approach. |
| title | A Two-Stage Architecture for NDA Analysis: LLM-based Segmentation and Transformer-based Clause Classification |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2603.09990 |