软件群体中基于交互序列的频繁模式挖掘算法研究

Research on Frequent Pattern Mining Algorithm Based on Interactive Sequences in Software Groups

互联网模式下,软件之间的交互日益频繁,具有交互特征的软件逐渐形成群体,其个体行为不断地影响着群体的特性.频繁模式挖掘是发现关联规则的一种重要技术,软件交互行为频繁模式对软件的安全性稳定性分析起着至关重要的作用.因此,从软件群体交互的角度提出了一种频繁模式挖掘算法SG-FIP.首先,定义了软件群体,构建基于时间段滑动的交互序列挖掘模型;其次,定义了一种基于时间段及交互次数的序列权重,提升了交互模式的兴趣度.第三,定义了约束系数,在预定义的滑动时间段内挖掘得到针对性更强、实时性更高的交互模式.最后,实验结果验证了交互模型的可行性,SG-FIP算法的时间消耗在一定程度上优于先前算法,并且精简了先前算法的挖掘结果,提高了挖掘精度,高效地挖掘出了软件群体中的频繁模式.

Software can run not only independently but also often collaboratively to perform tasks thus form larger group of software networks. The individual behavior of software constantly affects the characteristics of groups. Frequent interaction patterns play an important role in the security and stability analysis of software. Therefore,a frequent pattern mining algorithm called SG-FIP based on interaction sequences is proposed from the perspective of interaction software group. Firstly,the software group is defined,and an interaction sequence mining model based on the sliding time period is built. Secondly,a sequence weight based on a time period and interacting quantity is defined to enhance the interest of patterns. Thirdly,the constraint coefficient is defined to restrain mining results. In a predefined sliding time period,more targeted interaction sequence pattern is found by SG-FIP algorithm. Finally,experiments are simulated to verify the feasibility of interaction model. Time-consuming of SG-FIP algorithm is to some extent better than the previous algorithms,and inappropriate mining results of previous algorithms are filtered out to improve the mining accuracy.