摘要
[目的/意义]基金项目文本蕴含着丰富的前沿主题信息,通过细粒度识别科学研究前沿中的新兴主题,可以预测分析新兴主题未来发展趋势并可视化展示。[方法/过程]文章提出一种基于时间序列分析和SVM模型的基金项目新兴主题趋势预测与可视化分析方法,分析基金项目内外部特征属性,构建基金项目新兴主题探测公式,利用时间序列分析和支持向量机等深度学习算法模型,对新兴主题发展趋势进行预测分析,最后利用可视化分析软件Gephi进行可视化展示以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并结合专家咨询和传统论文聚类方法对比分析,表明该方法能够更加快速准确识别新兴主题,为我国科技政策制定提供决策支持和参考。
[Purpose/significance] Fund sponsored project documents contain rich information of frontier topics. Through identifying emerging topics of scientific research in fine-grained degree,this paper predicts and visualizes the future trends of emerging topics. [Method/process] The paper proposes a method of forecasting and visualizing the emerging topics of fund sponsored projects based on time series analysis and SVM model,in order to analyze the internal and external characteristics of fund sponsored projects,construct the emerging topics detection formula,and use deep learning algorithms such as time series analysis and SVM model to predict and analyze the development trend of new topics. Finally,visual analysis software Gephi is applied to visualize and reveal the competitive trend in the frontier area. [Result/conclusion] Through the experiment on graphene field data,and the comparison with expert consultation and traditional paper clustering analysis methods,it shows that the proposed method can identify the frontier topics more quickly and accurately,thus providing decision support and reference for national science and technology policy formulation.
出处
《情报理论与实践》
CSSCI
北大核心
2019年第1期118-123,152,共7页
Information Studies:Theory & Application
关键词
时间序列
基金项目
支持向量机模型
新兴主题探测
可视化
time series
fund sponsored project
Support Vector Machine model
emerging topic detection
visualization
