||
最近几年,每年在诺贝尔奖发榜之前,我们都会看到汤森路透的预测名单,但似乎准确率不太高。汤森路透预测时,一个很重要的依据是科学家发表论文的影响力(被引次数)。但是,在文献计量学中,有一个基本规则:影响≠质量,也就是说,影响力最大的论文的质量不见得就是最好的。
我们课题组选了一组高质量的论文(已获得诺贝尔奖的科研成果)和一组高影响的论文(在Web of Science数据库中被引次数很高),做了一项对照实验。借助引文曲线(被引用的历史过程)的分析框架,我们发现:高质量论文的被引曲线多数不规则,无法用曲线拟合的方法获取函数表达式;而高影响论文的被引曲线多数很规则,可以用曲线拟合的方法获取对数正态函数与指数函数表达式。
因此,从高影响论文中去寻找诺贝尔奖的潜在得主不符合历史规律。
论文发表在Malaysian Journal of Library & Information Science(SSCI收录),http://ejum.fsktm.um.edu.my/VolumeListing.aspx?JournalID=3 ,附论文标题与摘要:
A probe into the citation patterns of high-quality and high-impact publications
Abstract: Enlightened by Avramescu’s studies, we classified the citation patterns of high-quality and high-impact publicationsinto five mathematical types: lognormal-type, exponential-type, polynomial-type, wave-type and sleeping-beauty. Based on practical datasets of high-quality and high-impact publications, the five types of citation patterns are fitted well at statistical level. The results reveal that high-quality publications usually appear irregular citation curves (wave-type and sleeping-beauty) and high-impact publications tend to show regular ones (lognormal-type, exponential-type and polynomial-type).
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-5-3 09:19
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社