摘要
[目的/意义]从作者出发探测微观层面上人工智能领域研究前沿,并结合相关信息对前沿进行深入解读,以把握作者在整体中的研究态势、所处位置与特色,并发现基于不同属性识别出的研究前沿的类型与状态。[方法/过程]利用CiteSpace中的突变检测算法实现对高产作者突变术语的检测,再根据突变术语的原始来源论文抽取出对应突变术语的多种属性特征,进而依据属性从突变术语中提炼出研究前沿。[结果/结论]依据突变术语所属论文的题录属性,界定出作者层面的50个热点研究前沿、14个新兴研究前沿、34个成长研究前沿。高产作者的热点研究前沿代表领域内的研究热点与研究趋势,新兴研究前沿反映最近几年领域内比较新颖的研究技术或研究方法,成长研究前沿反映从新兴研究前沿到热点研究前沿的过渡状态。[局限]实验所涉及的高产作者数量受限,实验结论的普适性有待进一步验证;CiteSpace中突变检测的参数设置范围受软件本身所限制,无法灵活调整模型参数。
[Purpose/significance]To explore micro-level research frontiers in artificial intelligence from author-level,and combine related information for further interpretations.To grasp the author’s research situation,position and characteristics in the whole field.Also,to find types and states of research frontiers identified based on different attributes.[Method/process]We use the burst detection algorithm in CiteSpace to find burst terms of high yield authors,then extract various attributes according to original sources of burst terms.Finally,research frontiers are concluded from burst terms based on those attributes.[Result/conclusions]According to the bibliographic attributes of the papers of burst terms,50 hot research frontiers,14 emerging research frontiers and 34 growth research frontiers are defined at the author level.Hot burst terms represent research hotspots and research trends.Growing burst terms can predict research focus or research trends while emerging burst terms correspond to more granular research directions or technical means in a field.[Limitations]Limited by the relatively small number of high yield authors,the universality of the experimental conclusion needs to be further verified.The parameter setting range of burst detection in CiteSpace is limited by the software itself and cannot be adjusted flexibly.
出处
《情报理论与实践》
CSSCI
北大核心
2019年第9期22-27,共6页
Information Studies:Theory & Application
基金
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)
江苏省研究生科研与实践创新计划项目“面向知识创新服务的数据清洗实现模式研究”(项目编号:SJCX18_0135)的成果之一
