研发设计资源大规模领域本体构建方法  

Construction method of large-scale domain ontology of R&D design resources

作　　者：栗统 杜丽峰[2] 王磊[1] LI Tong;DU Lifeng;WANG Lei(School of Mechanical Engineering,Tianjin University,Tianjin 300350,China;Tianjin Tianduan Press Co.,Ltd.,Tianjin 300232,China)

机构地区：[1]天津大学机械工程学院,天津300350 [2]天津市天锻压力机有限公司,天津300232

出　　处：《锻压装备与制造技术》2022年第4期138-145,共8页China Metalforming Equipment & Manufacturing Technology

基　　金：国家重点研发计划项目(2018YFB1701801);天津市科技计划项目(18ZXRHGX00010)。

摘　　要：对企业中知识资源缺乏统一的组织与管理,共享效率低的问题进行了研究,提出了企业研发资源空间本体层中领域本体的构建方法,开展了本体构建技术研究。根据集团企业的需求将设计资源分类,并针对每类资源的非结构化文本数据运用Jieba分词进行分词以及停用词过滤、词性过滤等,通过LDA模型抽取本体中的概念,之后运用双语词典、基于word2vec的层次聚类、关联规则等算法进行关系识别,并在最后运用企业设计资源中软件资源的部分资料对本文提出的方法进行可行性验证。The problems of lack of unified organization and management of knowledge resources in enterprises as well as low sharing efficiency have been studied. The construction method of domain ontology in the ontology layer of enterprise R&D resources space has been proposed, and the research on ontology construction technology has been carried out. The design resources have been classified according to the needs of the group enterprise. The Jieba word segmentation has been adopted for the unstructured text data of each type of resources to perform word segmentation, stop word filtering, part-of-speech filtering, etc. The concepts in the ontology has been extracted through the LDA model while the relationships have been identified on the basis of bilingual dictionary, word2vec’s hierarchical clustering, association rules and other algorithms. Finally,some data of software resources in enterprise design resources has been adopted to verify the feasibility of the method proposed in the text.

关 键 词：研发设计资源 本体构建 概念抽取 关系识别 

分 类 号：TH166[机械工程—机械制造及自动化] TP274[自动化与计算机技术—检测技术与自动化装置]