基于结构-功能语义关联的技术机会识别研究  

Technology Opportunity Recognition Based on Structural-Functional Semantic Association

作　　者：张金柱[1] 叶晓宇 ZHANG Jinzhu;YE Xiaoyu(School of Economics and Management,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学经济管理学院信息管理系,江苏南京210094

出　　处：《情报科学》2024年第8期164-173,共10页Information Science

基　　金：国家自然科学基金面上项目“基于专利多模态内容和交易数据的互补技术识别与挖掘研究”(72374103);“基于表示学习的专利信息语义融合与深度挖掘研究”(71974095)

摘　　要：【目的/意义】当前研究主要从主题识别及演化和技术组合角度识别技术机会,尚需从更细粒度的技术要素间“结构-功能”关联角度提高准确性和结果可解释性。【方法/过程】首先基于实体关系抽取方法获取结构-功能技术要素实体以及实体间的多种关系,并对功能实体进行聚类整合;然后,基于结构-功能语义关联构建知识图谱,将其中与功能具有直接语义关联的结构和结构组合视为技术机会,并采用链路预测模型KG-Predict进行技术机会识别和结果评估;最后,通过获取技术机会在知识图谱中的链路来得到技术机会实现路径。【结果/结论】以扫地机器人领域为例,得到技术机会识别结果评估指标MRR为0.135,Hits@10为0.265,分别比未使用技术要素间语义关联的方法提高了0.064和0.123,证明了本文方法的有效性。此外,根据技术机会实现路径得到了技术机会详细解释。【创新/局限】从结构-功能视角识别技术机会,结合知识图谱得到技术机会实现路径。后续可继续细化结构关系并选用更好的模型提高准确性。【Purpose/significance】The current research mainly identifies technological opportunities from the perspective of topic recognition and evolution and technology combination,but it is still needed to improve the accuracy and interpretability of results from the perspective of more granular"structure-function"correlation among technological elements.【Method/process】In this paper,based on the entity relationship extraction method,the structure-functional technical element entities and the various relationships between entities are obtained,and the functional entities are clustered and integrated.Then,this paper constructs a knowledge graph based on structure-function semantic association,and regards the structure and structure combination with direct semantic association with function as technical opportunities,and uses the link prediction model KG-Predict to identify technical opportunities and evaluate the results.Finally,the path of technology opportunity realization is obtained by obtaining the link of the technology opportunity in the knowledge graph.【Result/conclusion】This paper uses patent data in the field of vacuum cleaning robot for empirical analysis,and the experimental results show that the MRR evaluation index for technological opportunity identification is 0.135,and Hits@10 is 0.265,which are 0.064 and 0.123 higher than that of the method without semantic correlation between technical elements,proving the effectiveness of the research method in this paper.In addition,the technology opportunities are explained in detail according to the path to realization.【Innovation/limitation】Identify technology opportunities from the perspective of structure and function,and obtain the realization path of technology opportunities based on the knowledge graph.Further studies can further refine the structural relationship and select better models to improve accuracy.

关 键 词：技术机会识别 实体关系抽取 知识图谱 链路预测 技术要素 

分 类 号：G254[文化科学—图书馆学]