摘要
命名实体识别在识别工业设备故障方面发挥关键作用,有助于故障预测、维护管理和智能决策。针对工业设备故障数据中存在的嵌套结构和长跨度问题,提出一种边界感知的实体识别方法。该方法通过边界感知精准定位实体跨距,并结合类别预测判断实体跨距的所属类别,以提高识别性能。此外,为解决标注数据的缺乏的问题,还构建面向工业设备故障的实体识别数据集。实验结果证明了该方法在工业设备故障实体识别方面的有效性,并为后续数据分析和知识图谱的构建提供了坚实基础。
Named Entity Recognition plays a key role in identifying industrial equipment faults,aiding in fault prediction,maintenance management,and intelligent decision-making.Aiming at the nested structures and long spans in industrial equipment fault,this paper proposes a boundary-aware entity recognition method.This method accurately located the span of entities through boundary detection and enhanced recognition performance by combining category prediction to determine the entity span's category.To tackle the scarcity of labeled data,this paper constructed an entity recognition dataset targeted at industrial equipment faults.Experimental results demonstrate the effectiveness of this method in recognizing entities related to industrial equipment faults,which lays a solid foundation for subsequent data analysis and knowledge graph construction.
作者
葛卫京
刘晓丽
杜亚峰
Ge Weijing;Liu Xiaoli;Du Yafeng(School of Mechanical Engineering,Shangqiu Institute of Technology,Shangqiu 476000,Henan,China)
出处
《计算机应用与软件》
北大核心
2024年第6期237-242,249,共7页
Computer Applications and Software
关键词
命名实体识别
预训练语言模型
工业设备
故障信息
Named entity recognition
Pre-trained language models
Industrial equipment
Fault information