Tag
This paper presents a multitask GLiNER-based framework for scalable monitoring of dataset usage in research literature, using synthetic data generation and LLM-based revalidation to address challenges in extraction, relation identification, and usage classification.