高关注论文视角下研究前沿识别多方法融合研究

Research on Multi-method Fusion of Research Front Identification from the Perspective of High-Profile Papers

[目的/意义]引文分析法与Altmetrics分别代表了不同侧重点下的学术评价方法,为聚焦研究前沿来源文献的遴选范围,将结合二者用于论文的综合影响力评价。不同的研究前沿识别方法优缺点不同,为更全面、客观地识别,将在4种不同的研究前沿识别方法得到的结果基础上,探索融合不同前沿识别结果的途径。[方法/过程]选取领域的高关注量文献集进行共被引分析、耦合分析,分别筛选出高关注、高被引文献集和高关注、低被引文献集进行词频分析,并在混合这两种文献集的基础上进行共词分析,对比不同研究前沿识别方法所得结果的异同,之后采用改进的线性加权融合模型进行前沿结果融合。[结果/结论]通过对人工智能领域进行实证分析,验证了本文方法的可行性。研究发现,耦合分析比共被引分析能识别到更多的研究前沿,二者识别到的前沿结果中有约40%的论文相同,与Jarneving的研究结果相近;词频分析法识别结果的特点是多且碎片化,共词分析识别结果的特点是少但更明确。融合后得到的识别结果包括人工智能算法与应用两类,其中,虚拟现实、中风康复、人脸识别、智能安防、智能医疗是时下最具潜力的人工智能应用领域。

[Purpose/significance]Citation analysis method and Altmetrics respectively represent academic evaluation methods in different ways.In order to select the source of frontier documents precisely,the two methods will be combined for the comprehensive impact evaluation of papers.Different research front identification methods have different advantages and disadvantages.In order to identify more comprehensively and objectively,first we use four different research front identification methods to get different identification results,then try to explore the way to integrate the identification results of different research front methods.[Method/process]Co-citation analysis and coupling analysis are carried out on the basis of high-attention documents,and word frequency analysis is carried out respectively on the basis of high-attention,high-cited documents and high-attention,low-cited documents.Besides,co-word analysis is carried out on the basis of the previous two document collections.After that,the similarities and differences of the results obtained by different research front recognition methods are compared.Then,the improved linear weighted fusion model is used to merge the frontier results.[Result/conclusion]Through empirical analysis in the field of artificial intelligence,the feasibility and effectiveness of this method are verified.The study found that coupling analysis could identify more research fronts than co-citation analysis.About 40%of the research fronts identified by the two methods are the same,which is similar to Jarneving’s research results.The characteristics of the recognition results of word frequency analysis are diverse and fragmented,the characteristics of coword analysis are fewer but more clear.The research front recognition results obtained by the fusion analysis method include two aspects,which are the algorithms and applications of artificial intelligence.Among them,virtual reality,stroke rehabilitation,face recognition,intelligent security,and intelligent medical care are the most potential artificial intelligence application fields.