DALDALL: Data Augmentation for Lexical and Semantic Diverse in Legal Domain by leveraging LLM-Persona

1. 1Leverage LLMs to generate persona-based synthetic data
2. 2Apply domain-specific strategies to prioritize quality over quantity
3. 3Use the generated data to augment existing legal datasets
4. 4Evaluate the performance of legal IR models using the augmented dataset

NLP researchers and legal domain experts can benefit from this framework to improve the quality and diversity of their datasets, and ML engineers can apply it to develop more accurate legal IR models

💡 Domain-specific data augmentation strategies can improve the quality and diversity of legal datasets

💡 Improve legal IR with DALDALL, a persona-based data augmentation framework using LLMs!