面向煤矿领域的文本关系抽取关键技术研究

Research on Key Technologies of Text Relation Extraction for Coal Mine

关系抽取是信息抽取的子任务,将关系抽取应用到煤矿的规范、章程等诸多复杂的文本信息方面,对于煤矿行业知识图谱的构建等研究有重要的价值。文章将目前基于规则、基于机器学习和基于深度学习的关系抽取等主要技术的方法和思路进行分析,并提出了使用BiLSTM-ATT(双向长短期记忆网络-注意力机制)模型来实现煤矿行业文本信息中实体关系的抽取。该课题可以为从事煤矿行业的人员和其他领域的研究提供较大的实际意义。

Relationship extraction is a sub-task of information extraction.Applying relationship extraction to many complex textual information such as coal mine specifications and regulations is of great value for the construction of knowledge maps in the coal mine industry.The article analyzes the current methods and ideas of rule-based,machine-based and deep-learning-based rela⁃tionship extraction,and proposes the use of BiLSTM-ATT(bidirectional long-short-term memory network-attention mechanism)model to implement coal mine industry texts.Extraction of entity relationships in information.This topic can provide greater practi⁃cal significance for the personnel engaged in the coal mine industry and other fields of research.