一种基于非独立同分布下K-means算法的系统日志分析方法  

System log analysis method based on K-means algorithm withinnon-independent and identical distribution

作　　者：谢青青 XIE Qingqing(Shandong Open University,Jinan 250100,China)

机构地区：[1]山东开放大学,山东济南250100

出　　处：《无线互联科技》2024年第21期94-99,共6页Wireless Internet Science and Technology

摘　　要：系统日志作为记录系统操作和事件信息的重要资源,对保障系统安全和优化系统性能具有至关重要的作用。利用K-means算法进行系统日志分析能够帮助管理员对日志进行分类管理,通过对相似日志条目的自动聚类,提高日志检索和管理的效率。传统K-means聚类算法一般采用欧氏距离作为相似性度量方法,该方法忽略了对象属性之间存在的耦合关系,是假设数据具有独立同分布的特性的,然而在现实的数据中,对象属性之间会存在一些复杂的耦合关系,是非独立同分布的。文章提出一种基于非独立同分布下K-means算法的系统日志分析方法,以非独立同分布的思想进行相似性度量。实验结果表明该方法能够获得较高的准确率和较低的聚类执行时间。As an important resource for recording system operation and event information,system logs play a vital role in ensuring system security and optimizing system performance.The K-means algorithm can help administrators classify and manage logs,and improve the efficiency of log retrieval and management through automatic clustering of similar log entries.The traditional K-means clustering algorithm generally uses Euclidean distance as a similarity measurement method,which ignores the coupling relationship between object attributes,and assumes that the data has the characteristics of independent and identical distribution,but in the real data,there will be some complex coupling relationships between object attributes,which are non-independent and identically distributed.In this paper,a system log analysis method for K-means algorithm within non-independent identical distribution is proposed,and the similarity is measured by the idea of non-independent identical distribution.Experimental results show that the K-means algorithm based on non-independent identical distribution proposed in this paper can obtain high accuracy and low clustering execution time.

关 键 词：非独立同分布 K-MEANS算法 日志分析 相似性度量 耦合关系 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]