摘要
针对转向架设计加工知识内容复杂、形式不统一、知识重用困难等问题,开展了基于图谱处理转向架知识的可行性研究。对收集的转向架知识基于数据类型特征进行分类,通过特征编码方式对离散型数据进行处理,利用实体识别和依存句法分析对文本型数据进行分解抽取与重组,以三元组形式建立文本型数据的知识图谱表达,并对离散型数据和文本型数据进行组合,建立了转向架知识的Neo4j图谱数据库,实现了转向架设计与加工中大量文本信息与离散信息的处理及高效重用,验证了利用知识图谱处理转向架知识的可行性。
In view of the problems of complex knowledge content,inconsistent form and difficulty of knowledge reuse in the process of bogie design and manufacturing,the feasibility study about bogie knowledge processing was carried out based on knowledge map.The knowledge of bogie was collected and classified based on the characteristics of data type.The discrete-type data were processed by feature coding,and the text data was decomposed and reorganized through the way by entity recognition and dependency syntactic parsing.The knowledge map representation of text data was established in triplet form.The Neo4j map database of bogie knowledge was established by combining the discrete-type data with the text data.The processing and efficient reuse of large amounts of text and discrete-type information in bogie design and manufacturing were realized,and the feasibility of processing bogie knowledge with the knowledge map was verified.
作者
马付建
李锡伟
黄文丽
MA Fujian;LI Xiwei;HUANG Wenli(School of Mechanical Engineering,Dalian Jiaotong University,Dalian 116028,China)
出处
《大连交通大学学报》
CAS
2023年第5期79-85,共7页
Journal of Dalian Jiaotong University
基金
2020年工业互联网创新发展项目
辽宁省自然科学基金项目(2021-MS-295)
辽宁省教育厅科学研究计划项目(JDL2020015)。
关键词
转向架
特征编码
依存句法分析
知识图谱
数据库
bogie
feature coding
dependency syntactic parsing
knowledge map
database
