基于Spark平台的恶意软件最大频繁子图挖掘方法  

Maximum frequent subgraph mining method for malware based on Spark

作　　者：周显春[1] 肖衡[1] 焦萍萍 邹琴琴 Zhou Xianchun;Xiao Heng;Jiao Pingping;Zou Qinqin(School of Information and Intelligence Engineering,Sanya University,Sanya 572022,China)

机构地区：[1]三亚学院信息与智能工程学院,三亚572022

出　　处：《现代计算机》2023年第14期57-61,共5页Modern Computer

基　　金：海南省自然科学基金项目资助(620MS064);三亚市院地科技合作项目资助(2019YD26);三亚学院优势专业建设项目(SYJZUS202203);三亚学院一流本科专业特色建设资助项目(SYJZZZ202212)。

摘　　要：为了解决传统子图挖掘算法时效性差的问题,设计了一种基于Spark平台的恶意软件最大频繁子图挖掘方法。该方法在保证挖掘信息完整的前提下,避免了挖掘所有频繁子图,采用了改进的FSMBUS方法来挖掘恶意软件的最大频繁子图,利用分布式架构Spark迭代计算优势,提高了挖掘效率。此外,改进算法还被应用于恶意软件同源性判定,改善了恶意软件检测效果。最后,通过对比实验结果,论证了该方法的高效性和可行性。In order to solve the problem of poor timeliness in traditional subgraph mining algorithms,a malicious software maximum frequent subgraph mining method based on the Spark platform was designed.This method avoids mining all frequent subgraphs while ensuring the integrity of mining information.It adopts an improved FSMBUS method to mine the maximum frequent subgraphs of malicious software,utilizing the advantages of distributed architecture Spark iterative computation to improve mining efficiency.In addition,the improved algorithm has also been applied to malware homology determination,improving the effectiveness of malware detection.Finally,by comparing the experimental results,the efficiency and feasibility of this method were demonstrated.

关 键 词：恶意软件 最大频繁子图 任务并行化 SPARK 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] O157.5[自动化与计算机技术—计算机科学与技术]