摘要
当前的数字档案分类方法存在测试集与训练集的基数之间差值偏大的问题,导致分类结果出现误差。为此,设计一种新的基于改进支持向量机的数字档案多标签分类算法。根据SVM函数确定数字档案信息的标签隶属度。求解深度指标,完成基于改进支持向量机的数字档案标签挖掘。获取数字档案信息样本,计算标签参量之间的相似度水平,实现基于改进支持向量机的数字档案多标签分类算法的设计。实验结果表明,研究方法下的标签测试集基数与训练集基数之间的差值始终小于350个,不会造成严重的数字档案信息错误分类问题。
Due to the large difference between the cardinality of test set and training set in the current digital file classification method,the classification results appear errors.Therefore,a new multi⁃label classification algorithm based on improved support vector machine is designed.The label membership degree of digital archive information is determined by SVM function.The depth index is solved and the label mining of digital archives based on improved support vector machine is completed.The information samples of digital archives are obtained,and the similarity level between label parameters is calculated.The multi⁃label classification algorithm of digital archives based on improved support vector machine is designed.Experimental results show that the difference between label test set cardinality and training set cardinality is always less than 350,which will not cause serious digital file information misclassification problem.
作者
张岚
张向阳
王金柯
杨铁军
刘骞
ZHANG Lan;ZHANG Xiangyang;WANG Jinke;YANG Tiejun;LIU Qian(Marketing Service Center(Measuring Center)of State Grid Henan Electric Power Company,Zhengzhou 450000,China)
出处
《电子设计工程》
2024年第3期41-44,49,共5页
Electronic Design Engineering
关键词
改进支持向量机
数字档案
多标签分类
容错系数
相似度
样本基数
improved Support Vector Machine
digital archives
multi⁃label classification
fault tolerance coefficient
similarity
sample base