期刊论文下载次数的概率密度服从对数正态分布  

Probability density of downloaded times for periodical articles obey log-normal distribution

作　　者：毛国敏[1] 吴何珍[1] 生冬梅[2] 

机构地区：[1]中国地震局地球物理研究所,北京100081 [2]中国地震局工程力学研究所,黑龙江哈尔滨150080

出　　处：《辽宁师专学报（自然科学版）》2017年第3期100-105,共6页Journal of Liaoning Normal College(Natural Science Edition)

摘　　要：许多社科类指标的概率密度分布都不是正态的,这就为有效合理利用这些指标做进一步分析带来了困难.当指标数据分布为非中心对称时,一般不能直接利用原始数据做分析,需要对数据的分布有所了解,在此基础上做必要的变换,才能做有关的统计分析.期刊论文下载次数是一个非常客观的指标,可显示论文被使用和受重视的程度,随着期刊检索系统的发展,获取期刊论文被下载的数据已成为可能.在数据基础上,选用对数正态函数模型,研究期刊论文下载的分布问题.对不同学科类别、载文规模和学术影响力的3种期刊,运用非线性迭代计算方法模拟,试图寻找一种简单的确实能反映期刊论文下载次数指标分布的内在规律.得出期刊论文下载指标的概率密度服从对数正态分布,这种简单的对数正态分布规律有可能具有普适性,为今后合理有效地利用期刊计量指标提供了参考依据.The probability density distribution of many social science indexes is not normal,which makes it difficult to make further analysis on the effective and rational utilization of these indexes.The analyses cannot be based on the original data if all these indexes are non-centrosymmetry.It need to know the distribution of data and make the necessary transformation,and then to do the statistical analysis.Downloaded indicators for periodical articles are very objective to show the used and valued papers.It will be possible to obtain the downloaded data with the development of the periodical index system.Based on the data,this paper selects a log-normal function model to study the distribution of the download of periodical papers.By using nonlinear iteration simulation method to find a simple inner rules which can show downloaded indicator distribution for periodical articles through different subject classification,load scale and academic influence.It is concluded that the probability density of the download indicators for periodical articles obey log-normal distribution which is universal to use as a reference for the reasonable and effective use of periodical measurement indexes in the future.

关 键 词：科技期刊 论文下载次数 概率密度分布 对数正态函数 模拟 

分 类 号：G252.8[文化科学—图书馆学]