基于改进Eclat算法的资源池节点异常模式挖掘  

Mining anomaly pattern of nodes in resource pool based on improved Eclat algorithm

作　　者：高强[1] 张凤荔[1] 陈学勤[1] 王馨云 耿贞伟 周帆[1] 

机构地区：[1]电子科技大学信息与软件工程学院,成都610054 [2]云南电网有限责任公司信息中心,昆明650217

出　　处：《计算机应用研究》2018年第2期333-338,共6页Application Research of Computers

基　　金：国家自然科学基金资助项目(61602097;61272527);四川省科技支撑计划资助项目(2016GZ0065;2016GZ0063);中央高校基本科研业务费资助项目(ZYGX2015J072)

摘　　要：云计算环境中,资源池节点异常模式挖掘对于快速诊断节点状态具有重要作用。针对云环境下计算资源池、存储资源池、网络资源池节点数据特征,对资源池节点状态信息进行预处理,利用关联规则算法挖掘资源池节点参数状态信息之间的关联关系,如高位—高位和低位—高位模式等。提出了一种基于限制属性连接并具有垂直数据格式的关联规则算法i-Eclat算法。i-Eclat算法通过转换资源池节点状态数据格式、建立非频繁2-项集以减少连接次数,并构建信息存储结构体来限制冗余属性连接。实验表明,所提出的方法可以有效发现云计算资源池节点之间的隐藏关系;同时,i-Eclat比经典算法计算性能更优,特别是针对较大数据集的处理。In cloud computing, mining anomaly pattern of nodes plays an important role in promptly diagnosing node states of resource pool. According to the data characters of computing resource, storage resource and network resource, it preprocessed the information of node status. And it used association rule algorithm to mine the rules among the nodes＇ parameter status of re- source pool, e. g. , ＂high level-high level＂ and ＂low level-high level＂. This paper proposed an improved Eclat algorithm, called i-Eclat, which based on vertical data format of restrictive attribute connection, i-Eclat reduced the number of connections by transforming the format of state data and constructing a non-frequent 2-itemsets, as well as it constructed the storage structure to limit the redundant attribute connections. Extensive experiments have been conducted to demonstrate that this method can find the hiding rules among resource pool＇ s nodes. And it also shows that i-Eclat outperforms traditional methods on computing efficiency, especially on processing big data sets.

关 键 词：模式异常挖掘 关联规则 资源池 i-Eclat算法 云计算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]