一种基于TDT模型的基金项目科学研究前沿识别方法研究

Research on Scientific Research Fronts Detection Method of Fund Sponsored Projects Based on TDT Model

[目的/意义]由于传统以论文分析数据源的科学研究前沿方法存在主题探测的时滞性及方法的局限性,需要更加准确识别出蕴含更多前瞻价值信息的基金项目数据中的研究前沿主题,细粒度识别科学研究前沿主题类型。[方法/过程]提出一种基于TDT模型的基金项目科学研究前沿识别方法,借鉴TDT模型中多要素融合分析及归一化处理的思想,分析基金项目的资助强度、时间维度、主题维度等项目特征属性,构建基金项目科学研究前沿探测公式,识别出热门研究前沿主题、新兴前沿主题及未来前沿主题3种科学研究前沿主题,从而揭示前沿领域竞争态势。[结果/结论]通过以石墨烯领域数据进行实验,表明该方法能够更加快速准确地前瞻识别出科学研究前沿主题,弥补单一主题维度进行前沿识别的不足,更好地进行主题发展预测分析。

[Purpose/significance] Traditional scientific research fronts detection methods taking analytic data of papers as data sources have time lag in topic identification and limited methods. This paper aims to detect the topics of research fronts in fund sponsored projects more accurately,which contains more prospective and valuable information, and to refine the topic types.[Method/process] The scientific research fronts detection methods of fund sponsored projects based on the TDT model is proposed. Based on the multi-factor integration analysis and normalization idea in TDT model,the paper analyzes the funding intensity,time dimension,topic dimension,and other attributes of fund sponsored projects,constructs the formula of scientific research fronts detection,and identifies the hot research front topics,the emerging front topics and future front topics,so as to reveal the competitive situation of scientific research fronts. [Result/conclusion]Through the experimental data in the field of graphene,the results show that this method can identify topics of scientific research fronts more quickly and accurately,which makes up the limitations of fronts detection within a single dimension,and can better predict and analyze topic development.