情报科学 ›› 2021, Vol. 39 ›› Issue (3): 179-184.

• 博士论坛 • 上一篇    下一篇

基于作者特征和期刊特征的学术论文被引频次
预测模型构建与分析

  

  • 出版日期:2021-03-01 发布日期:2021-03-15

  • Online:2021-03-01 Published:2021-03-15

摘要:

【目的/意义】预测学术论文的被引频次不仅能够帮助作者提前了解个人研究成果未来的应用价值,还能够
帮助期刊编辑人员预估期刊的影响力。【方法/过程】通过分析论文作者特征、论文所在期刊特征与论文被引频次的
相关性,发现影响论文被引频次的重要因素;利用逐步回归分析建立用于预测论文被引频次的数学模型,用于预测
论文未来几年内的被引频次。【结果/结论】实验发现与学术论文被引频次呈显著相关的10个因素;发现可用期刊即
年指标和作者之前发表论文的篇均被引频次的线性组合来预测学术论文未来的被引频次。【创新/局限】分析预测
结果,发现可通过以下途径提高预测模型的准确度:将某学科的所有期刊限定到学科内某一特定期刊、将作者之前
发表论文的篇均被引频次限定为以第一作者身份发表论文的篇均被引频次、将论文的研究内容纳入预测模型等。

Abstract:

【Purpose/significance】In order to help the author understand the future value of personal research in advance and help
journal editors predict the impact of journals, we propose the method for predicting citation frequency of academic papers.【Method/
process】In this paper, we find that the important factors affecting the journal paper cited frequency by analyzing the correlation be⁃
tween the author characteristics and papers cited, periodical characteristics and papers cited; and establish the mathematical model
for forecasting the journal paper cited frequency by stepwise regression analysis.【Result/conclusion】This experiment found 10 fac⁃
tors which significantly correlated with the cited frequency of journal papers; the future cited frequency of journal papers can be pre⁃
dicted using the linear combination about the periodical immediacy index and the average cited frequency per paper.【Innovation/limi⁃
tation】The prediction model still has errors even it has a certain coincide with true value. Finally, we provide the effective way to im⁃
prove the accuracy of prediction by analyzing the causes of error.