中国人民大学【求是论坛】 | GLAM领域中文化遗产资源的智能计算与创新应用

Tang Xuemei shared her research on relation extraction from classical Chinese historical documents. Starting from the needs of historical studies and leveraging text clustering methods, the research proposed a knowledge representation model tailored for ancient historical texts to guide entity-relation annotation and dataset construction. To address the scarcity of annotated data in this domain, a domain-specific entity-relation dataset was built, effectively mitigating the data shortage challenge in relation extraction tasks. Furthermore, to tackle few-shot and long-tail distribution problems, the study introduced a model-collaboration-based framework for relation extraction and developed a platform for knowledge graph generation. By integrating theories and methods from information science, history, and deep learning, this research offers new insights and practical approaches for entity-relation extraction in other types of classical texts.

May 24, 2025 9:00 AM — 12:00 PM

Knowledge Representation and Relation Extraction from Classical Chinese Historical Documents

Online