摘要
[目的/意义]从时序主题演化的角度,构建探测科学领域研究前沿的途径与方法,为科技创新和科研决策提供有效支撑。[方法/过程]提出基于时序主题关联演化的前沿探测三阶段模型。首先将领域文本集合按照时间窗进行划分,利用LDA主题模型生成各个时间窗的研究主题;再通过相邻时间窗主题之间的相似度建立主题关联,设置主题关联过滤规则并对无效主题关联进行剔除;最后,按主题之间的关联关系构建主题演化路径,根据主题路径变化探测科学领域研究前沿。[结果/结论]以石墨烯领域中文科技文献为研究案例,进行时序主题关联演化分析,探测出石墨烯纳米复合材料及其应用、石墨烯电极材料研究以及石墨烯光电性能和应用三大研究前沿,验证了方法模型的有效性。
[Purpose/significance]According to the sequential evolution of research topics,methods of research frontier detection in scientific field are constructed,which could be helpful to provide effective support for scientific and technological innovation and scientific research decision-making.[Method/process]A three-stage model of frontier detection is proposed based on sequential topic association evolution.Firstly,the domain text set is divided according to the time windows,the research topics in each time window are generated by using the LDA topic model;then the topic associations are established by the similarity of adjacent time windows’topics,and the topic association filtering rules are set to eliminate the invalid topic associations;finally,the topic evolution paths are built according to the topic association relationships,and their changes can be used to detect the research frontiers in the scientific field.[Result/conclusion]Taking the Chinese scientific literature in the field of graphene as a research case,the sequential topic association analysis is utilized and three research frontiers have been detected,including graphene nanocomposites and their applications,graphene electrode materials,photoelectric properties and applications of graphene,which verified the validity of the proposed model.
出处
《情报理论与实践》
CSSCI
北大核心
2019年第7期144-150,共7页
Information Studies:Theory & Application
基金
江苏省社会科学基金项目“领域知识分析视角下文献知识关联揭示及应用研究”(项目编号:17TQB009)
国家社会科学基金重大项目“面向知识创新服务的数据科学理论与方法研究”(项目编号:16ZDA224)
江苏省2011社会公共安全科技协同创新中心的成果
关键词
时序主题演化
科学前沿
相似度
LDA主题模型
sequential topic evolution
scientific frontier
similarity
LDA topic model
