摘要
[目的/意义]基于文献计量学的新兴技术预测方法能够避免现有德尔菲法的专家依赖性,但是文献所属技术领域的划分仍需依据领域知识设计复杂的检索式。[方法/过程]提出了一种基于机器学习和时间序列预测的新兴技术预测方法,LDA主题模型与SVM(支持向量机)分类模型连用的机器学习方法解决了海量论文摘要数据按照新兴技术类别的多分类,通过ARIMA(差分自回归移动平均模型)模型对时序论文未来数量的预测,分析以技术为驱动力的新兴技术发展趋势。[结果/结论]最后,以机器人技术为例,提取Web of Science(WOS)论文数据库上万条数据,对7个新兴领域发展趋势进行预测,为新兴技术预测工作提供有益的数据和方法支持。
of Paper Classification
Dong Fang1, Liu Yufei2 ,Zhou Yuan3 （ 1. School of Mechanical ＆ Engineering, Huazhong University of Science and Technology, Wuhan 430074 ; 2. School of Life Science and Technology, Huazhong University of Science and Technology, Wuhan 430074 ; 3. School of Public Policy and Management, Tsinghua University, Beijing 100084） Abstract [ Purpose/Sgnificance] Emerging Technologies prediction method based on bibliometric method can avoid Delphi experts de- pendence, however the technology field that the literature should be categorized into is still based on the design of complex search queries according to the field knowledge. [ Method/Process ] In this paper, a new prediction method based on machine learning and time series forecasting is proposed. The machine learning method combining the LDA theme model and SVM （ Support Vector Machine） classifica- tion model is used to solve the problem of multi-class abstract of paper classification in emerging technology. Besides, applying the ARI- MA （Autoregressive Integrated Moving Average Model） model to predicting paper number in the future can contribute to the analysis of technology-driven emerging technology trends. [ Result/Conclusion] In the end, this paper takes the robot technology as an example to extract the relative data from the Web of Science （WOS） Database, with the purposes of forecasting the developing trend of seven specific emerging fields and then providing effective data and methods to support the further learning in prediction of the whole emerging technolo-gies.
出处
《情报杂志》
CSSCI
北大核心
2017年第7期40-45,133,共7页
Journal of Intelligence
基金
国家自然科学基金"支持技术预见的多源异构大数据融合与时序文本预测方法研究"(编号:91646102)研究成果之一
国家自然科学基金"面向2035的中国工程科技发展路线图绘制理论与方法研究"(编号:L1624045)研究成果之一
国家自然科学基金"面向2035的中国工程科技发展路线图应用案例及软件研究"(编号:L1624041)研究成果之一
国家自然科学基金"2035发展战略文献计量与专利分析方法研究"(编号:L1524015)研究成果之一
国家自然科学基金"示范项目对产业技术轨道的作用机理研究--以新能源汽车和新能源产业为例"(编号:71203117)研究成果之一
