摘要
基于支持向量机学习模型,使用万方期刊数据库中医学、卫生的有关标题和摘要数据,对医学、卫生大类下的R7中的9个小类进行了自动分类研究。在中文信息处理知识和技术的基础上选取分类特征,在分类过程中主要采取了基于低密度多特征的训练方法。在互信息、卡方统计、交叉熵和证据权值4个不同的统计特征量的开放测试中,自动分类的查全率和准确率都取得了相对令人满意的结果。
Based on the SVM（Support Vector Machine）learning model,and by the use of the data of the titles and abstracts concerning traditional Chinese medicine and sanitation in Wanfang Periodical Database,this paper makes an automatic classification study of 9 sub-classes in R7 of medicine and sanitation class.The classification features are selected on the basis of the Chinese information processing knowledge and technology.During the classification process,the paper mainly uses the training method based on low density and multi-feature.The recall and pertinency ratios of automatic classification is relatively satisfactory in the open test of 4 different statistic features,that is,mutual information,chi-square statistics,cross entropy and proof weight.
出处
《情报理论与实践》
CSSCI
北大核心
2011年第4期115-118,共4页
Information Studies:Theory & Application
基金
教育部人文社会科学重点研究基地重大项目"基于智能信息处理的知识挖掘技术及应用研究"的研究成果之一
项目批准号:08JJD870225
