Construction of Knowledge Graph for Geological Data in Xinjiang Collection
To further enhance the integrated utilization of geological data information in the Xinjiang Natural Resourc-es Archives and break the current limitation of archival retrieval only through catalog search,a knowledge graph is intro-duced to optimize the management of geological materials in the Xinjiang Archives.Partial geological materials in the ar-chives are used as the data source,and entities and relationships are determined through ontology construction.The Xinji-ang geological materials data is manually annotated using a BIO sequence labeling method.The BERT-BiLSTM-CRF model is employed for knowledge extraction,and the Neo4j graph database is used to store the knowledge of Xinjiang geological materials,completing the construction of the Xinjiang Geological Materials Knowledge Graph.Experimental results show that the BERT-BiLSTM-CRF model achieves an accuracy rate of 98.1777％and an F1 score of 97.8921％,significantly outperforming the BERT-CRF,BERT-IDCNN-CRF,and BERT-BiGRU-CRF models.The construction of the Xin-jiang Geological Materials Knowledge Graph can provide a foundation for the development of a"Digital Archives"in the Xinjiang Natural Resources Archives and enhance the socialization of Xinjiang geological data big data services.
NETL
NSTL
万方数据
