基于密度分布单类支持向量机的科技文献分类研究  

Scientific Literature Classification Research Based on the Density Distribution of OCSVM

作　　者：董微[1] 赵捷[1] DONG Wei;ZHAO Jie(Institute of Scientific and Technical Information of China,Beijing 100038,China)

机构地区：[1]中国科学技术信息研究所,北京100038

出　　处：《情报工程》2018年第3期67-72,共6页Technology Intelligence Engineering

基　　金：科技部科技创新战略研究专项"基础研究领域年度重点创新进展报告"(ZLY201636);NSTL专项基金项目"开放学术资源建设"(2016XM16)

摘　　要：在OCSVM单分类问题上,科技文献自动分类时交叉学科的分类并未得到良好的解决,且支持向量构造的超平面未考虑到非支持向量的影响,本文提出了一种基于密度分类的单类支持向量机的分类算法,将支持向量的密度分布引入目标函数。实验结果表明,该算法能够较好的将交叉学科的科技文献进行主题分类。The problem of the scientific literature automatic classification on cross subject in the One-Class SVMs is still not solved,and the small portion of samples called support vectors fully decide the hyperplane,whereas all the nonsupport vectors have no influence on the hyperplane.This study proposed a classification based on the density of one class support vector machines classification algorithm.Density distribution of support vectors is introduced into the objective function.The experimental results show that the proposed algorithm can classify the interdisciplinary scientific literature subject better.

关 键 词：科技文献 文本分类 密度分布 单类支持向量机 

分 类 号：G250.7[文化科学—图书馆学]