学术论文影响力的四维测度：引文综合评估、社交网络评议、同行直接评价和期刊间接体现

There are four measures (i.e. citation comprehensive evaluation, social networks appraisal, direct peer review, and indirect based-journal assessment) can be used to determine the academic impact of single publications.The latter three categories hold the characteristics of quick action,but the citation analysis is not the same. Many pioneers of citation analysis such as Garfield have contributed greatly to the advancement of the method first proposed by Gross et al in 1927. However, it has been criticized frequently that the inherited limitation of citation analysis as a retrospective measurement is the time lags. As a contrast to the citation-based measurements, the dominant feature of the other three measures is quick action, which can provide the alternative, or a timely assessment. However, are these alternative new measurements valid and meaningful assessments of article’ influences in comparison with the widely-accepted traditional retrospective measurements? If some of these measurements do provide reliable assessments, how can we integrate them with the well-established citation analysis to derive at a better assessment model? These problems are very interesting and the study on them are in the ascendant. This proposed research aims to analyze multiple data sources to investigate the validity and applicability of these measurements and those traditional evaluations based on citation, and to derive a new four-dimensional assessment model of article assessments with reasonable weights in the digital age. We should set forth its theoretical soundness. Specifically, data will be pulled from multiple sources, such as Faculty of 1000, Web of Science, PLOS, etc. This data-intensive project is challenging but will have implications for dynamically selecting diverse meaningful measurements as an integral part of assessments of an publication’s academic influences. Further, the special methods proposed by this research, such as the four-dimensional assessment model, the comparative analysis of 10 kinds of mixing patterns, using the citation ranking’s annual correlation analysis to establish the time period, the shift of annual weights,the design of benchmarking papers in peer review, the real-time evaluation based on news outlets and so on, will promote the growth and development of the knowledge of single publication evaluation.

引文综合评估、社交网络评议、同行直接评价和期刊间接体现可以分别从不同角度测度学术论文的影响力。与第一类方式相比，后三类都具有即时性的特征，但第一类引文分析方式自1927年由Gross等学者提出后，经Garfield等大批先驱的不懈努力获得了飞速发展，但始终无法摆脱天生的滞后性。因此，如何有效地将四者结合起来进行学术评估是一个很有研究价值的新问题，国内外在此方面的探索方兴未艾。本课题将在对比分析四类评估方式各种具体做法特点的基础上，建立四位一体评估模型研究不同指标多种结合模式在理论上的合理性，并借助于F1000、WoS、PLOS等资料库中大量数据研究各种结合模式的应用价值。本课题提出的四位一体评估模型、十种结合模式对比分析、权重此消彼长的变动趋势分析、引用排名年度之间相关性分析及相关系数聚类分析、基于标杆论文判断法、媒体报道为主的即时评价法等都是较为独特的做法，必将促进该领域相关知识的发展。

本课题在深入探讨期刊评价、引文评价、网络评价、同行评议四类指标特点的基础上，提出了多种全新的评价学术论文影响力的方法，并借助于WoS、F1000、领域代表性期刊和学科顶级期刊等资料库中的大量数据研究各种评价模式和评价指标的理论和应用价值，同时还探讨了对学术论文影响力进行综合评价和多角度评价的可行方案。本课题取得的重要成果有：（1）提出全新的“编辑团队学术指数”，使用图情领域21种知名期刊的738名编辑团队成员的数据进行评估，研究表明新指数与期刊声誉有显著的正相关关系。（2）首次系统地研究合并arXiv预印本和期刊版本文章引文计数的理论与方法。为了将研究成果两种版本的引用计数整合起来，提出了四个可行的方案和一个通用方法。（3）提出全新评价指标“贡献加权引用指数”用于评价学术论文影响力。基于380多万篇论文的研究表明，本研究提出的三个具体评价指标在识别论文影响力方面表现出不同的特征。（4）利用232名编委会成员的信息可以将77种图情期刊分为五类，即MIS（22刊）、IS（14刊）、LS（28刊）、SM（3刊）和“其他类”（10刊）。（5）基于1933名商业领域学者的研究表明，在职业生涯早期发表顶级期刊论文的研究人员，后续表现确实更好一些。（6）基于7000多篇量子纠缠实验领域的论文，研究发现A.Zeilinger和潘建伟是该领域最具优势的科学家；机构中维也纳大学表现最佳，中国科学技术大学也不错。（7）以生命科学类四份顶刊（即Cell,JAMA,Lancet,NEJM）上发表的论文为例，分析了F1000Prime推荐与否与引用次数高低之间的关系，发现不同期刊呈现出不同的特点。（8）以46份代表性期刊上的13000篇左右论文为基础，全面分析了我国管理科学与工程学科的研究进展、研究热点及演化趋势。（9）在对“大团队发展和小团队颠覆科学技术”这篇Nature封面论文质疑的基础上，提出全新的评价学术论文颠覆性的系列指标。本课题一定程度上弥补了以往研究的不足，不仅提出了一些在学术评估领域颇具创新性的概念和方法，而且对相关问题进行了多角度分析，对学术论文影响力的测算提出了新的研究思路，拓展了学术评价领域的研究主题和内容，推进了学术评估理论和实践的完善和发展。