基于网络表示学习的本体语义挖掘与功能语义检索方法  

Ontology semantic mining and functional semantic retrieval method based on network representation learning

作　　者：倪子健 李文强[1,2] 唐忠[1,2] NI Zi-jian;LI Wen-qiang;TANG Zhong(School of Mechanical Engineering,Sichuan University,Chengdu 610065,China;Innovation Method and Creative Design Key Laboratory of Sichuan Province,Sichuan University,Chengdu 610065,China)

机构地区：[1]四川大学机械工程学院,四川成都610065 [2]四川大学创新方法与创新设计四川省重点实验室,四川成都610065

出　　处：《工程设计学报》2021年第5期539-547,共9页Chinese Journal of Engineering Design

基　　金：国家自然科学基金资助项目(52075350);四川大学宜宾市校战略合作专项(2020CDYB-37)。

摘　　要：为充分利用本体概念之间隐含的语义关系,以支持产品功能创新设计过程,提出了一种基于网络表示学习的本体语义挖掘与功能语义检索方法。首先,基于本体中确定的语义关系,利用网络表示学习挖掘隐含的语义关系;然后,基于语义类比的向量运算,建立本体概念之间潜在的功能语义关系,并对功能语义向量进行表达;最后,通过功能语义向量的相似度计算实现由用户功能需求向跨领域功能性设计资料的扩展,并建立相应的设计资料检索方法和流程。产品设计示例表明,所提出的本体语义挖掘与功能语义检索方法有利于从产品功能角度获取跨领域设计知识,可为设计人员提供更多的灵感。In order to make full use of the implicit semantic relationship between ontology concepts to support the product function innovation design process, an ontology semantic mining and functional semantic retrieval method based on the network representation learning was proposed. Firstly, based on the determinate semantic relationships in the ontology, the network representation learning was used to mine the implicit semantic relationships;then, based on the vector operation of semantic analogy, the potential functional semantic relationship between ontology concepts was established, and the functional semantic vector was expressed;finally, through the similarity calculation of the functional semantic vector, the expansion from functional requirements of users to cross-domain functional design data was realized, and the corresponding design data retrieval method and its process were established. The product design example shows that the proposed ontology semantic mining and functional semantic retrieval method is conducive to obtain cross-domain design knowledge from the perspective of product function, which can provide more inspiration for designers.

关 键 词：功能设计 网络表示学习 语义挖掘 向量表达 检索 

分 类 号：TH122[机械工程—机械设计及理论] TP391[自动化与计算机技术—计算机应用技术]