基于多源文本挖掘的技术演化路径识别  

Technology Evolution Path Recognition Based on Multi-source Text Mining

作　　者：刘怀兰[1] 刘盛 周源[2] 张睿 Liu Huailan

机构地区：[1]华中科技大学机械科学与工程学院,湖北武汉430074 [2]清华大学公共管理学院,北京100084

出　　处：《情报理论与实践》2022年第11期178-187,共10页Information Studies:Theory & Application

基　　金：国家自然科学基金项目“面向2035的高档数控机床技术路线图研究”的成果,项目编号:L1824043。

摘　　要：[目的/意义]提出一种基于文档向量化和自动化短语挖掘的改进主题建模方法(Doc2Vec-KMeans++-TopMine,DKT),从多维视角识别领域内的技术演化路径,展示领域内技术的发展与现状。[方法/过程]获取研究领域内的论文、专利、产品等多源数据,综合采用文档向量化、聚类算法和短语挖掘算法,完成领域多源数据的融合主题挖掘,通过语义相似度计算进行多源融合主题的关联与演化路径的识别,采用“主题河流图”可视化方法绘制多维度技术演化路径,从前沿研究、技术研发、市场应用等多维度视角出发开展领域演化分析。[结果/结论]选取数控机床领域进行实证研究,实验结果表明,利用DKT方法挖掘到的演化路径信息丰富,可以进行多维技术演化路径分析,并证明了其有效性,同时还得到了多个维度视角下的领域发展过程,发现了数控机床领域的3种技术演化模式。[Purpose/significance]This paper proposes an improved topic modeling method(Doc2Vec-KMeans++-TopMine,DKT)based on document vectorization and automated phrase mining to identify the technological evolution path in the field and demonstrate the development of technology in the field.[Method/process]This paper obtains multi-source data such as papers,patents,products in the research field,comprehensively uses document vectorization,clustering algorithms and phrase mining algorithms to complete the fusion topic mining of multi-source data in the area.Furthermore,the semantic similarity calculation is used to associate the multi-source fusion topics and identify the evolution path.Finally,the visualization method of“Topic River”is used to draw the multi-dimensional technology evolution path,carrying out the domain evolution analysis from the multi-dimensional perspectives of frontier research,technology research and development,and market application.[Result/conclusion]This paper selects the field of CNC machine tools for empirical research.The experimental results show that the evolution path information mined by this method is rich,which can be used for multi-dimensional technology evolution path analysis.At the same time,the effectiveness of this method is proved.This paper also obtains the field development process from the perspective of multiple dimensions,and finds three technological evolution modes in the field of CNC machine tools.

关 键 词：技术演化路径 多源数据 主题建模 文本挖掘 Doc2Vec TopMine 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]