Beyond Raw Text: Knowledge-Augmented Italian Relation Extraction with Large Language

Relation extraction (RE) is a fundamental NLP task that identifies semantic relationships between entities in text, serving as the foundation for applications such as knowledge graph completion and question answering. In real-world deployments, organizations frequently encounter low-resource scenarios where labeled training data is scarce, making effective RE particularly challenging. Existing approaches often rely on external knowledge sources to augment training data, but such resources can be noisy, incomplete, or misleading for model learning. To address this limitation, we propose an approach that leverages the reasoning capabilities of Large Language Models (LLMs) to generate reliable background knowledge for RE tasks on Italian texts