【作者】 张旭；

【导师】 成颖；

【作者基本信息】 南京大学 ， 情报学， 2021， 硕士

【摘要】 学术期刊作为学术成果发表的重要载体,是当下科学研究以及教育体系的重要组成部分,在个人评价、职位晋升、项目申请、机构评价等评价活动中扮演着重要角色。自学术期刊诞生以来,围绕学术期刊评价开展的研究已相当丰富。目前,期刊评价方法可分为两类,即基于定性分析的同行评议法以及基于定量分析的文献计量法。同行评议法通常选取特定领域的专家对其领域内的期刊质量进行评估,是当下期刊评价的主流方法。不过,同行评议也存在诸如主观性较强、专家知识储备不足等较为明显的缺陷。当前的研究主要从技术和制度方面着手,改进同行评议本身的不足,以提升其评价的效果和效率。文献计量作为一种重要的评价手段,可以解决同行评议主观性强等不足,是辅助同行评议必要且重要的手段。不过,主流影响力评价指标,如期刊影响因子、h指数等也存在诸如忽略引用行为差异,将每一次引用等同对待的不足,为此,学界基于复杂网络理论,采用Page Rank算法以及该算法的衍生指标(如特征因子、声望指标等)对期刊进行了评价研究。值得注意的是,无论是影响力,还是特征因子等指标,都未能有效结合同行评议的专家意见。现今,学界公认的最优的评价方法是有机结合同行评议和文献计量优势的混合评价方法。据此,本文采用了一种可结合专家意见的Page Rank衍生算法——Trust Rank开展期刊的权威性评价研究。尽管不同专家对不同期刊的排序存在意见不一致的情况,但是对学科一流期刊的认定鲜有异议,本研究在采用Trust Rank算法时,基于学术界普遍认可的一流期刊,通过期刊引用网络传递期刊TR值,从而达成对所有期刊的评价。基于此,本文以管理学期刊作为研究对象,基于CSSCI(2000-2018)的引文数据开展期刊权威性的评价研究。此外,本文还拟从多角度、多方位对期刊的全貌进行评价。创新作为科学研究的第一动力,评估研究成果的创新性不仅有助力国家占领科技新高地,还能为科研发展提供新方向。故此,本研究引入颠覆因子测度期刊的颠覆性创新。而影响力作为反映科研成果扩散表现的重要测量维度,评估期刊的影响力对科学发展有强大的推进作用,基于影响力评价的研究已有许多,如期刊影响因子等。本研究拟观测影响力(期刊影响因子)与权威性(Trust Rank)、创新性(颠覆因子)等维度在期刊评价时的协同效应,此外,本研究还进一步探讨了不同维度的测度指标在评价时存在的时间窗以及指标融合等问题。本文的主要研究结论有:(1)Trust Rank算法作为Page Rank算法的衍生算法,在期刊评价上的表现优异,这种融合专家意见的算法不仅在评价上优于Page Rank,还与期刊影响因子具有较高的相关性。其次,该算法具有一定的稳健性,评价效果优异,可广泛用于期刊评价。(2)颠覆因子作为评估研究颠覆性的有效指标,在本文的CSSCI数据集上表现不佳,区分度较低,说明该指标不适用于基于CSSCI数据的图书馆、情报与文献学专业期刊的颠覆性评估。(3)利用不同指标进行组合评价时,引文时间窗可根据不同指标的表现选取最优值,而无需将不同指标的时间窗设置成同一值。(4)多指标评价时,若不同指标融合以后的评价效果更优,可根据评价需求对指标进行融合,从而简化评价流程。(5)相较于单一属性的评价效果,从多个视角对期刊进行评价,评价效果更为真实,更能全面地反映出期刊的全貌。更多还原

【Abstract】 As an important carrier of academic publications,academic journals are an important part of the current scientific research as well as education system,and play an important role in academic evaluation such as personal evaluation,job promotion,project application,and institutional evaluation.Since the emergence of academic journals,there has been a considerable amount of research conducted around academic journal evaluation.Currently,journal evaluation methods can be divided into two categories,namely peer review based on qualitative analysis and bibliometrics based on quantitative analysis.The peer review is the mainstream method of journal evaluation,as experts in a particular field are usually selected to assess the quality of journals in their field.However,peer review has some obvious drawbacks,such as a high degree of subjectivity and a lack of expert knowledge.Current research focuses on technical and institutional aspects to improve the shortcomings of peer review itself in order to enhance its effectiveness and efficiency.As an important evaluation tool,bibliometrics can address the shortcomings of peer review and is a important tool to assist peer review.However,mainstream impact evaluation metrics,such as journal impact factor and hindex,also suffer from shortcomings such as ignoring differences in citation.For this reason,based on complex network theory,academic community has conducted research on the evaluation of journals with the Page Rank algorithm and the derivatives of this algorithm(e.g.eigenfactor,SJR,etc.).It is noteworthy that neither the impact factors,nor the eigenfactor and other metrics cannot effectively combine with expert opinion from peer review.Nowadays,it is accepted that the optimal evaluation method is a hybrid evaluation method that organically combines the advantages of peer review and bibliometrics.Accordingly,this paper uses Trust Rank,a Page Rank-derived algorithm which combines expert opinion,to evaluate the authority of journals.Although there is disagreement between different experts on the ranking of different journals,they reach an agreement on the identification of top journals in the discipline.In this study,the Trust Rank algorithm is used to reach an evaluation of all journals based on the top journals that are generally recognised by the academic community,and the journal TR values are passed through the journal citation network.This paper takes management science journals as the research object and conducts a study on the evaluation of journal authority based on the citation data of CSSCI(2000-2018).In addition,this paper also intends to evaluate the full picture of journals from multiple perspectives.As innovation is the first driving force of scientific research,assessing the innovativeness of research results not only has the potential to help the country capture new heights in science and technology,but also can provide new perspective for scientific research development.Therefore,this study introduces the disruption index to measure the disruptive innovation of journals.As an important measure of the proliferation of research results,assessing the impact of journals plays a powerful role in promoting scientific development.This study aims to observe the synergistic effects of impact(Journal Impact Factor)with dimensions such as authority(Trust Rank)and innovation(Disruption Index)in the evaluation of journals.In addition,this study further explores the citation window and the integration of indicators in the evaluation of different dimensional measures.There are five main findings of this paper.Firstly,Trust Rank algorithm,as a derivative of Page Rank algorithm,has excellent performance in journal evaluation.This algorithm incorporating expert opinions not only outperforms Page Rank in evaluation,but also has high correlation with journal impact factor.The algorithm also has certain robustness and excellent evaluation effect,which can be widely used in journal evaluation.Secondly,the disruptive index,as an effective indicator for assessing the disruption of research,performs poorly on the CSSCI dataset in this paper,with low differentiation,indicating that the indicator is not applicable to the disruption assessment of professional journals in the field of library,information and science based on CSSCI data.Thirdly,when using different indicators for combined evaluation,the citation window can be selected as the optimal value according to the performance of different indicators.It’s unnecessary to set the same value for different indicators’ citation windows.Fourthly,when evaluating multiple indicators,the evaluation process can be simplified by fusing the indicators according to the evaluation needs if the evaluation effect is better after fusing different indicators.Finally,compared with the evaluation effect of a single attribute,the evaluation effect of journals from multiple perspectives is more realistic and can reflect the whole picture of journals more comprehensively.更多还原