基于国家自然科学基金成果海量数据的统计学方法研究

As the coming of the big data era, mass data processing of the fund achievements has become feasible in technology, but on the methodology research is far from mature. This project focuses on statistical methods based on the science fund data. First establish SCI papers, citation database, keywords and hot word database, citation network database between disciplines ,that produced by fund assistance, on this basis,to put forward efficient keyword extraction method and hot tracking method, carries on the science fund tendency and the link between the international research hot spot analysis; Based on the time series model, establish dynamic intervention model,and analysis the intervention of the causal effect of different corresponding time series model because of the trend and strength of scientific fund assistant every year ; Establish interaction diagram paper quoted status - between disciplines influence network to describe and quantify interdisciplinary influence on other disciplines. This project by building a unified framework for comprehensive evaluation of science foundation funding to provide methodology basis for funding effect.

随着大数据时代的到来，基金成果的海量数据处理在技术上已变得可行，然而在方法论研究上还远不够成熟。本项目重点研究基于科学基金成果数据的统计学方法。首先建立基金资助所产生的SCI论文及引文数据库、关键词和热点词数据库、学科间论文引用网络数据库，以此为基础，提出高效的关键词提取方法和热点追踪方法，进行科学基金资助倾向与国际科研热点之间的关联分析；以时间序列模型为基础，拟建立动态干预模型，用以分析由于每年科学基金资助倾向和力度的不同对相应的时间序列模型产生的干预的因果效应；建立学科间论文引用情况的交互图——学科影响网络，用来描述和量化各学科对其它学科的影响力。本项目通过构建一个统一的框架为综合评价科学基金对各学科的资助成效提供方法论依据。