[目的/意义] 归集重大科研项目数据资源、探索数据的有效利用以揭示其蕴含的前瞻科技前沿主题,对于以“创新驱动,情报先行”为科技部门、创新企业做好“耳目、尖兵、参谋”,助力科学规划部署、角逐科技创新、抢占创新制高点有着及其重要的意义。[方法/过程] 以科技创新型国家重大科研项目数据资源归集为基础,通过实证从项目所属科研计划、项目名称及关键词、项目三个方面探索基于科研项目数据的科技前沿识别有效路径。[结果/结论]科研计划、项目名称、关键词、等一定程度上揭示项目内容的数据字段在科技前沿识别方面可以发挥作用,但存在数据字段缺失、数据规范性不足、杂质较多,识别方法工具待改进等问题,并提出以全球科技创新型国家重大科研项目数据库为基础的宏观、中观、微观组合式前沿识别路径。

[Purpose/Significance] This paper is written to summarize the data resource of major scientific research projects and explore the effective utilization of data, so as to reveal the included forward-looking themes of frontier tech. It is of great significance for being “detector, pioneer and adviser” for the “innovation-driven and intelligence-prioritized” technological departments and innovative enterprises, supporting scientific planning and arrangement, chasing after technological innovation and striving for innovation superiority. [Methodology/Process] Based on the resources gathering of major national scientific projects in the technological innovation-oriented countries, the paper will explore the effective paths of identifying frontier tech according to the data of scientific research projects from perspectives of scientific research plans that projects are included, project name and key words as well as project brief. The method used is empirical investigation. [Results/Conclusions] To some extent, scientific research plan, project name, key words and briefs reveal that, the data fields of projects can work on identifying frontier tech. But there are still problems like absence of data fields, lack of data standardization, much irrelevant information and identification methods and tools to be improved. Besides, this paper puts forward a macroscopic, midscopic and microscopic-combined path of frontier identification, based on the major scientific research projects databases of worldwide technological innovation-oriented countries.

中的情报学项目数据为基础,通过人工判读以内容分析法和对比分析法对项目的主题分布、热点分布、研究侧面、研究背景等方面进行分析5；白如江等(2017)通过对科技规划文本和数据文本两种不同数据源中蕴含的科学研究前沿主题相似度对比分析,采用新兴度和热点度两个指标,识别出新兴研究前沿主题和热点研究主题两类科学研究前沿主题,进而揭示出前沿领域竞争态势6；孙晓玲等(2017)利用我国国家自然科学基金资助项目的论文成果用基于层次 Dirichlet过程模型和主题相似度映射算法,分析研究前沿主题的时序演化7；王效岳等(2017)以 NSF 资助的碳纳米管研究领域数据为研究对象,利用 PLDA 模型以及主题的资助时间、资助金额和中心性指标进行研究前沿主题探测8。