Study on The Metallogenic Regularity of Rare Metal and REE Deposits in Daxinganling Region Based on Knowledge Graphs

This study focuses on the rare metal and rare earth element (REE) deposits in Daxing'anling region, aiming to construct a knowledge graph(KG) using deep learning and natural language processing techniques to support the computation and reasoning of geological data. The research collected and analyzed 342 articles related to rare metal and REE deposits in Daxing'anling region, published on CNKI between 1997 and 2024. The Bert-BiLSTM-CRF model was used for named entity recognition, identifying geological entities such as deposit names and wall rock alterations, while the Att-BiLSTM model was employed to automatically extract relationships between entities. The processed knowledge was imported into the Neo4j graph database, enabling the visualization of the KG. This study validated the effectiveness and accuracy of the intelligent construction of the KG through experiments. The results indicate that the KG construction can effectively integrate multi-source geological data, help in understanding regional geological patterns, and provide a tool for geoscientists and decision-makers to quickly process and analyze geological data. The KG has broad application prospects.