作　　者： ;

机构地区： 广东工业大学华立学院

出　　处： 《智能计算机与应用》 2019年第3期266-268,272,共4页

摘　　要： 为了提高对用户行为特征挖掘能力,需要对用户行为特征多维度文本数据进行优化聚类处理,提出一种基于联合半监督学习的大数据聚类算法。采用分段线性拟合方法进行用户行为特征大数据线性规划处理,提取用户行为特征大数据的互信息特征量,结合联合关联规则检测方法进行用户行为特征多维度文本数据的统计分析,构建大数据分布的关联属性样本集,采用联合半监督学习分类器进行数据分类,结合多传感量化跟踪识别方法进行聚类中心自动搜索,提高聚类收敛性。仿真结果表明,采用该方法进行用户行为特征多维度文本数据聚类处理的信息融合性能较好,数据聚类中心的自动搜索能力较强,提高了大数据分类检索能力。 In order to improve the ability of user behavior feature mining,it is necessary to optimize the clustering of user behavior feature multi-dimensional text data.A big data clustering algorithm based on joint semi-supervised learning is proposed.The piecewise linear fitting method is used to deal with the user behavior feature big data,and the mutual information feature quantity of user behavior feature big data is extracted.Combined with the joint association rule detection method,the multi-dimensional text data of user behavior characteristics are analyzed,and the association attribute sample set distributed by big data is constructed,and the joint semi-supervised learning classifier is used to classify the data.The clustering center is automatically searched by multi-sensor quantization tracking and identification method to improve the clustering convergence.The simulation results show that this method has better information fusion performance and better automatic searching ability of data clustering center,which improves the ability of big data classification and retrieval.

关 键 词： 联合半监督学习 大数据 用户行为特征 聚类

领　　域： []