文章摘要
薛陕,董诚,韩红旗,张均胜,高雄,王力.科研主题演化中三种典型社区发现算法对比研究——以植物甾醇信号为例[J].高技术通讯(中文),2021,31(11):1164~1172
科研主题演化中三种典型社区发现算法对比研究——以植物甾醇信号为例
Comparison of three typical community discovery algorithms in topic evolution research——a case study of Brassinosteroids research areas
  
DOI:10.3772/j.issn.1002-0470.2021.11.006
中文关键词: 科研主题演化;社区发现算法;植物油菜素甾醇;关键词共现网络
英文关键词: evolution of the research topic, community discovery algorithm, Brassinosteroids, keyword co-occurrence network
基金项目:
作者单位
薛陕  
董诚  
韩红旗  
张均胜  
高雄  
王力  
摘要点击次数: 801
全文下载次数: 560
中文摘要:
      科研主题识别和演化分析是了解科研领域发展现状和趋势的重要手段。近年来,基于复杂网络和网络演化计量的社区发现算法成为揭示科研主题演化的主要方法,有必要深入探讨该类算法在主题识别和演化分析中的优点与不足。本文选取植物油菜素甾醇领域,构建了关键词共现网络,对比了Newman MM、Ball Overlapping和Blondel这3种典型社区发现算法的科研主题识别和演化追踪结果,并结合专家知识对3种算法效果进行分析。实验结果表明,Blondel算法在科研主题的识别和演化追踪研究中运算速度最快,识别的主题更为准确,能够较好反映领域主题演化。该研究从算法适用性角度为科研主题演化追踪研究提供了参考。
英文摘要:
      Identification and analysis of the evolution state of scientific research topics are important means to understand the current situation and trend of research topics. In recent years, community discovery algorithm based on complex networks and network evolution metrics has become one of the main methods to reveal scientific research topics. Therefore, it is necessary to explore the advantages and disadvantages of these algorithms in topic discovery and evolution research. Brassinosteroids hormone research area is selected as an example, the keyword co-occurrence network in this field is constructed, and the performances of three typical community discovery algorithms (Newman MM, Ball Overlapping and Blondel) in scientific research topic recognition are compared; the accuracy of the detection of research topic and the evolutionary analysis results is judged by expert knowledge. The results show that the Blondel algorithm has great advantages in the identification and evolution tracking of scientific research topics. It has not only the fastest speed of operation, but also the more accurate topic recognition. It provides a reference for the evolution tracking of scientific research topics from the perspective of algorithm applicability.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮