Cross-Domain Data Selection and Augmentation for Automatic Compliance Detection
Researchers have developed a new method for improving the accuracy of automated compliance detection systems. The study focuses on cross-domain data selection and augmentation, addressing the challenge that models trained on one set of regulations often perform poorly on others. By employing strategies like random sampling, cross-entropy difference, importance weighting, and embedding-based retrieval, the team demonstrated that targeted data selection significantly reduces negative transfer, paving the way for more reliable and scalable compliance automation across diverse legal texts. AI
IMPACT Improves cross-domain generalization for NLP models, potentially enhancing automated legal compliance tools.