基于深度集成学习的社交网络异常数据挖掘算法  

Anomaly data mining algorithm in social network based on deep integrated learning

作　　者：戴礼灿 代翔 崔莹 魏永超 Li-can DAI;Xiang DAI;Ying CUI;Yong-chao WEI(The Second Laboratory,Southwest China Institute of Electronic Technology,Chengdu 610036,China;Scientific Research Office,Civil Aviation Flight University of China,Guanghan 618307,China)

机构地区：[1]西南电子技术研究所第二实验室,成都610036 [2]中国民用航空飞行学院科研处,四川广汉618307

出　　处：《吉林大学学报（工学版）》2022年第11期2712-2717,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(U19A2078);四川省科技计划重点研发项目(2020YFG0009)。

摘　　要：针对当前社交网络异常数据挖掘算法设计过程中未提取社交网络数据特征,导致异常数据检测率低、检测运行时间长的问题,提出一种基于深度集成学习的社交网络异常数据挖掘算法。该算法首先挖掘社交网络数据,依据数据挖掘结果提取社交网络数据特征;然后利用提取到的数据特征构建深度集成学习模型,通过该模型对异常数据进行预测,以此获得社交网络异常数据,实现社交网络异常数据挖掘。实验结果表明:通过对该算法进行异常数据检测率测试和异常数据检测运行时间测试,验证了该算法的准确性高、有效性强、实际应用效果好。In the current design process of social network abnormal data mining algorithm,social network data features are not extracted,resulting in low abnormal data detection rate and long detection running time. Therefore,a social network abnormal data mining algorithm based on deep integrated learning is designed. Mining social network data,extracting social network data features according to the data mining results,using the extracted data features to build a deep integrated learning model,using the model to predict abnormal data,so as to obtain abnormal social network data,and realize abnormal social network data mining. Experimental results show that the algorithm has high accuracy and effectiveness and good practical application effect through abnormal data detection rate test and abnormal data detection running time test.

关 键 词：深度集成学习 社交网络 异常数据挖掘 特征提取 

分 类 号：TP748[自动化与计算机技术—检测技术与自动化装置]