期刊引文概率分布模型的建立与应用  

Establishment of journal paper citation probability distribution model and its application

作　　者：霍速[1] 谭智敏[1] HUO Su;TAN Zhi-min(Library of Agriculture School,Jilin University,Changchun 130062,Jilin Province,China)

机构地区：[1]吉林大学图书馆

出　　处：《中华医学图书情报杂志》2019年第7期21-26,共6页Chinese Journal of Medical Library and Information Science

基　　金：2018年CALIS农学中心研究项目“涉农学科引文网络的建立、分析与应用”(2018028)

摘　　要：目的:分析引文网络中期刊论文引文数据分布,做好期刊文献推荐工作。方法:通过对吉林大学农学部发文的引文分布实例分析,验证其偏离幂律分布的特征,建立通用的引文概率分布模型并拟合数据。结果:通过类比引用行为和物理中的随机扩散建立的引文概率分布模型,可以很好地拟合偏离幂律分布特征的引文数据,并给出拟合参数,结合概率推荐算法建立了一个优化的文献推荐模型。结论:将期刊引文概率分布模型用于优化文献推送模型可增加当前文献推送的精度或减少当前推算法的计算量,降低当前各种高精度的优化算法计算量过大的限制。Objective To analyze the distribution of journal paper citation data in journal network in order to improve journal literature recommendation.Methods A general journal citation probability distribution model was established by analyzing the citation distribution of journal papers published by Jilin University Agriculture School.The characteristics of its deviation from the power-law were verified and the journal paper citation data were fitted.Results The journal citation probability distribution model was established by comparing the citation behaviors and random diffusion in physics,which can effectively fit the citation data of the characteristics of its deviation from the power-law and provide the fitting data.An optimized journal literature recommendation model was thus established according to the probability recommendation algorithm.Conclusion The accuracy of journal literature recommendation is increased while the calculation load of recommendation algorithm and the overload calculation of various optimized algorithms are reduced when the journal citation probability model is used in optimized journal literature recommendation model.

关 键 词：引文网络 幂律分布 扩散模型 文献推荐 

分 类 号：G252.8[文化科学—图书馆学] G237.5