多标签隐性知识显性化下的数据挖掘算法  

Data Mining Algorithm Based on Multi-Label Tacit Knowledge Dominance

作　　者：刘利民[1] 张勇 LIU Li-min;ZHANG Yong(Institute of Computer Engineering,Guilin University of Electronic Technology,Beihai Guangxi 536000,China;College of Medical Information Engineering,Chengdu University of TCM,Chengdu Sichuan 610075,China)

机构地区：[1]桂林电子科技大学计算机工程学院,广西北海536000 [2]成都中医药大学医学信息工程学院,四川成都610075

出　　处：《计算机仿真》2023年第4期504-508,共5页Computer Simulation

摘　　要：对多标签数据进行挖掘时,由于数据挖掘模式的差异,导致算法加速比较低。提出基于SECI模型与属性分类的多标签数据挖掘算法。应用SECI理论建立数据转化模型,将多标签数据的隐性知识显性化处理。结合Relief F算法和互信息,提取多标签数据特征。通过属性分类方法,按照类内距离平方和最小、类间距离平方和最大的原则设计多标签数据挖掘模式,获取数据挖掘结果。在MVVM模式的作用下,建立挖掘结果交互方案,获取实时数据挖掘结果。仿真结果表明:所提出的数据挖掘算法应用后,加速比得到了有效提升。At present,the speedup ratio of the algorithm is relatively low in mining multi-label data due to the difference in data mining modes.Therefore,a multi-label data mining algorithm based on SECI model and attribute classification was proposed.Firstly,SECI theory was applied to build a data transformation model externalizing the tacit knowledge of multi-label data.Secondly,the Relief F algorithm was combined with mutual information to extract multi-label data features.According to the principle of minimizing the square sum of intra-class distance and maximizing the square sum of inter-class distance,the attribute classification method was used to design a multi-label data mining mode to obtain data mining results.Under MVVM mode,an interactive scheme of mining results was established to obtain real-time data mining results.Simulation results show that the speedup ratio is effectively improved after using the proposed algorithm.

关 键 词：属性分类 多标签数据 数据挖掘 特征选择 隐性知识 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]