多源引文复合研究：方法比较与分布规律

发布时间：2018-01-05 23:32

  本文关键词：多源引文复合研究：方法比较与分布规律　出处：《华东师范大学》2017年硕士论文　论文类型：学位论文

  更多相关文章： Scopus WoS Google Scholar 多源数据 复合引文分析 分布规律

【摘要】：鉴于 Scopus、Web of Science(WoS)与 Google Scholar(GS)所收录的期刊在利用库内引文指标进行测评时,结果往往不尽相同,三个单源数据库在计量学研究和应用中各有局限,单纯利用任一数据库进行期刊、学者测评都可能导致信息缺失,为了减少评估特定学科领域的期刊影响力时的偏差,更为全面的思路是将三个数据库的数据结合计算,构成基于多个不同数据库来源数据的综合性引文分析测度。论文以细胞生物学、植物学、图书情报学、管理学、语言学和哲学这六个学科2009-2014年数据为实证基础,尝试对比Scopus、WoS和GS三个单源数据库的引文指标差异。结果发现,三个单源数据库的总被引数值结果不尽相同,各库中引文数据总体满足GSScopusWoS;排名越靠前的期刊,在各个数据库中的引文数据差异越大;由于各个数据库中引文数据的差异,导致期刊基于引文指标进行测评时,结果在各库间显著不同。在选择复合指标时,以期刊被引次数这一底层参量为切入点,采用可以反映整体数据的平均水平的算术均值作为基本复合方法,在此基础上补充体现数值分布均衡性的几何均值和调和均值,对各期刊进行多源数据复合引文研究。均值测度结果证明,复合型指标总体上能较为准确地反映期刊在学科中的地位;利用均值复合方法得到的最终指标值,其结果与Scopus期刊的被引次数最为接近,其数值结果总体上满足GSCA≥CMScopusWoS;多源数据复合引文与单个数据库引文的期刊排名差异呈现"两端小、中间大"之现象;累计均值(算术、几何、调和)与期刊数之间未证实存在布拉德福定律,但在期刊数量等同的三分区内,累计被引均值满足形如n2:1n:1的经验分布;算术均值、几何均值和调和均值虽有差异,但亦有线性关联,应用中可选其一。本研究还使用变异系数,对单源与多源数据的离散度进行了分析。在其他复合方法上,还探索了论文在多个数据库中的最大施引文献集合,作为在论文层次复合方法的补充。理论与实证结果均证明,采用均值法和最大施引文献法对WoS、Scopus和GS三个数据库进行的复合引文分析,均可在一定程度上均衡了三个数据库差异性,得到一个相对全面、公正的期刊测评参数,从而更为全面的反映计量学中的引文数据。在实际的评价应用中,若数据中极端值较少的情况下,算术均值、几何均值和调和均值三种复合方法可换用,若引文数据在三个数据库中差异较大,则采用几何均值和调和均值相较于算术均值的结果更佳。结合本文的理论研究内容,与我国目前科研评价中的问题与现状,本研究也提出针对科研评价中的期刊评价、学科差异、科研人员评价和评价方法运用四个方面的政策建议。
[Abstract]:In view of the fact that the journals included in Scopusa Web of Science WoS and Google Scholaria are evaluated by citation indexes in the database. The results are often different, the three single-source databases have their limitations in metrology research and application. Using any database for journals, scholars' evaluation may lead to the lack of information. In order to reduce the deviation in evaluating the journal influence of a specific subject area, a more comprehensive idea is to combine the data of the three databases. A comprehensive citation analysis measure based on multiple data sources from different databases is presented in this paper, including cell biology, botany, library and information science, and management. Based on the 2009-2014 data of six disciplines of linguistics and philosophy, this paper tries to compare the differences of citation indexes between Scopusus WoS and GS. The total citation values of the three single source databases are different, and the citation data in each database generally meet the requirements of GSScopus WoS; The higher the ranking of journals, the greater the differences in citation data in each database; Because of the difference of citation data in each database, the results of periodical evaluation based on citation index are significantly different from each other. When selecting compound index, the bottom parameter of periodical citation number is taken as the starting point. The arithmetic mean, which can reflect the average level of the whole data, is adopted as the basic composite method, and the geometric mean and harmonic mean, which reflect the equilibrium of numerical distribution, are supplemented on this basis. The multi-source data compound citation of each periodical is studied. The results of mean value measure prove that the composite index can accurately reflect the status of the periodical in the subject. The final index value obtained by the mean value composite method is the most close to the number of citations in Scopus journals, and the numerical results generally meet the requirements of GSCA 鈮，

本文编号：1385367