网络数据库访问路径关联特征分析查询优化  

Correlation Feature Analysis and Query Optimization of Network Database Access Path

作　　者：张延芳[1] 袁浩[1] 

机构地区：[1]郑州大学西亚斯国际学院,河南郑州451150

出　　处：《计算机仿真》2015年第9期437-440,共4页Computer Simulation

摘　　要：网络数据库具有动态和不确定性,对其有效访问可以提高对网络大数据的挖掘和数据查询能力。网络数据库的访问路径的关联特征分析是实现数据库优化查询的有效途径,传统的网络数据库访问路径关联特征分析方法采用平均互信息关联维提取算法,当关联维特征出现独立同分布状态时,提取性能受限。提出一种数据流相邻时间段模式匹配的网络数据库访问路径关联特征分析算法,实现对数据流查询的优化设计。根据点的位置以及数据块在各维上最大长度,判断是否存在一个包含该点的簇,在新的数据流到达系统时,触发事件并将参数数据插入参数窗口中保存。查询节点数据分发直接信任值为前一周期信任值衰减后与当前信任值的平均值,建立等值查询条件机制,进行数据流相邻时间段模式匹配,在属性值和用户检索机制中设定查询条件,进行特征关联。仿真结果表明,该算法在维度和数量级足够大的时,在Esper中应用流聚类算法在稳定性一致的基础上,网络数据库访问有明显的时间优势。数据访问的耗时较传统方法有所减少,展示了算法对数据库访问路径规划和数据查询的优越性能。In the paper, a new correlation feature analysis algorithm was proposed based on data flow pattern matching, to realize the optimal design of data stream query. According to the position of point and the maximum length of data blocks in each dimension, whether there exists a cluster containing the point was determined, and when new data stream arrives at the system, the events were triggered and the parameter data were inserted into the parame- ter window to save. The node data were queried to distribute the direct trust values which were the average values of previous cycle trust attenuation values and current trust values, and the equivalent query mechanism was established. The pattern matching with data flow in adjacent time was implemented, and the query condition was set in property value and user retrieval mechanism, to make feature association. Simulation results show that if the dimension and quantity are large enough, on the basis of the consistent stability of application stream clustering algorithm in Esper, network database access has obvious advantage in time by using this algorithm. The time consuming of data access has been reduced, and the superiority of the algorithm has demonstrated for the database access path planning and data query.

关 键 词：网络数据库 数据挖掘 访问 特征分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]