矿山知识图谱构建的实体关系抽取方法研究  

Research on Entity Relation Extraction Method of Mine Knowledge Graph Construction

作　　者：杨森森 李海涛[1] 杜伟升 薛珊珊 YANG Sensen;LI Haitao;DU Weisheng;XUE Shanshan(Chinese Institute of Coal Science,Beijing 100013,China)

机构地区：[1]煤炭科学研究总院,北京100013

出　　处：《煤炭技术》2024年第11期259-263,共5页Coal Technology

摘　　要：知识图谱描述实体及其关系,对煤矿管理、安全监控等提供支撑。构建煤矿知识图谱面临数据多样性和语义关系挑战。提出一种基于ALBERT-BiLSTM-Att-CRF的方法,通过特征提取、上下文捕捉、注意力机制和CRF层优化实体关系抽取,展示较高的准确率和F1值,为煤矿智能化提供技术支撑。Knowledge graph describes entities and their relationships,and provides support for coal mine management and safety monitoring.The construction of knowledge map of coal mine faces the challenge of data diversity and semantic relation.Proposes a method based on ALBERT-BiLSTM-AttCRF,which displays high accuracy and Fi value through feature extraction,context capture,attention mechanism and CRF layer optimization entity relationship extraction,and provides technical support for coal mine intelligence.

关 键 词：矿山 知识图谱 实体关系抽取 ALBERT-BiLSTM-Att-CRF模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]