论文部分内容阅读
利用CNKI引文数据库,以图情领域共19本期刊53243篇文献为统计数据源,从单篇论文、作者、期刊三种粒度,分别对文献下载频次与被引频次进行数据正态性检验、相关性分析及曲线估计,并探讨利用下载频次预测被引频次的可行性。实验表明,下载频次与被引频次的相关性在不同粒度下差异较大:单篇论文粒度下相关性不强,作者粒度下呈显著的二次函数正相关,而期刊粒度下呈显著的三次函数正相关。因此,从作者或期刊粒度,利用下载频次预测被引频次是可行的。
Using the CNKI citation database, 53243 articles of 19 periodicals in the picture area were used as the statistical data sources, and the data normality was tested on the download frequency and the cited frequency of the article from the three kinds of granularity of the single essay, author and periodical, respectively Sexual analysis and curve estimation, and discuss the feasibility of using downloaded frequency to predict cited frequency. Experiments show that the correlation between download frequency and citation frequency is quite different under different granularities: the correlation is not strong under the granularity of a single paper, and significant quadratic function is positively correlated with the author’s granularity, while the journal granularity is significantly three times Function is related. Therefore, it is feasible to use the download frequency to predict the citation frequency from the author or journal granularity.