Researchers have developed an ontology-driven framework to improve the accuracy of document-level relation extraction. They found that a significant source of noise in datasets comes from structural inconsistencies within relational triples, such as violations of ontology constraints and logical contradictions. By enforcing structural consistency during training, they were able to reduce these errors and improve model generalization performance. --- Why it matters: This matters because it highlights the importance of ensuring data quality in relation extraction tasks, which can have a significant impact on downstream applications that rely on accurate information extraction. Source: https://arxiv.org/abs/2608.20856