摘要
[目的/意义]从机构层面探测领域研究前沿,以把握机构在整体中的研究动向与状态,发现机构在研究中所处的位势、特色及其与整个研究领域切合程度。[方法/过程]运用突变检测算法识别机构的研究前沿;采用定性分析法探究机构的前沿变化;通过定量分析法和余弦相似度计算公式,探究机构与总体以及机构间研究前沿的相似程度。[结果/结论]以中国科学院和卡耐基梅隆大学为例,探测出前者前沿较侧重算法模型分析,后者较侧重智能应用研究;通过机构与总体前沿重合度分析,将高产机构划分为引领型、同步型、追赶型、均衡型和特色型;通过机构间前沿相似度分析,探测出特色型机构与其他机构前沿一致性较低,而国内机构间前沿相似度较高等特点。突变算法检测到的机构前沿能反映出机构整体的前沿变化,机构在整体中的位置以及机构间前沿的相似性程度,能够为探明机构研究方向、整体实力、所处地位及其差异提供参考。[局限]前沿突变术语清洗过程中,未处理语义相似的词汇。
[Purpose/significance]In order to grasp the research trends and status of organizations in the whole,we detect the frontiers from the institutional level,which could also help to find out the position,characteristics and research relevance of organizations in this filed.[Method/process]We take burst detective algorithm to identify the research frontiers of institutions,qualitative analysis to explore the frontier changes of institutions,and quantitative analysis and cosine similarity to research the status of institutions and the similarities among institutional research frontiers.[Result/conclusion]We find that the frontiers of Chinese Academy of Sciences are more related with algorithms and models,while that of Carnegie Mellon University mostly focus on application research;the productive institutions can be specified into five types:leading,synchronous,catch-up,balanced and characteristic institutions;the similarities of frontiers between characteristic institutions and others are relatively low,while that between domestic institutions are relatively high.The frontiers of institutions detected by the burst detective algorithm are effective,and can help to find out the research trends,position and differences among organizations.[Limitations]When cleaning burst terms,some synonymic vocabularies were not processed.
出处
《情报理论与实践》
CSSCI
北大核心
2019年第9期16-21,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”的成果之一,项目编号:16DZA224
关键词
科研机构
研究前沿
突变检测
人工智能
余弦相似度
scientific research institution
research front
burst detection
artificial intelligence
cosine similarity