基于社团检测算法的二进制程序模块化方法  

Binary Program Modularization Method Based on Community Detection Algorithm

作　　者：朱肖雅 杨启超 尹小康 刘胜利 ZHU Xiaoya;YANG Qichao;YIN Xiaokang;LIU Shengli(State Key Laboratory of Mathematical Engineering and Advanced Computing,Zhengzhou 450001,China)

机构地区：[1]数学工程与先进计算国家重点实验室,河南郑州450001

出　　处：《信息工程大学学报》2023年第5期607-613,共7页Journal of Information Engineering University

基　　金：科技委基础加强项目(2019-JCJQ-ZD-113)。

摘　　要：对大规模二进制程序进行模块化时,现有方法存在较大时间开销,基于软件系统的“模块”与复杂网络的“社团结构”具有一定的特征相似性,研究使用社团检测算法进行二进制程序模块化。目前存在大量的社团检测算法,为此提出一个社团检测算法选择框架,该框架包括3个选择标准以及4个评估标准。根据选择标准进行算法过滤,选出3种具有代表性的社团检测算法,并使用评估标准来评估算法的模块化效果。通过多组对比试验,从中选择了一种效果较好的社团检测算法Louvain用于大规模二进制程序的模块化。实验表明,Louvain在效果方面优于目前最优的模块化形式概念分析(Formal Concept Analysis,FCA)方法,且实验平均耗时仅仅为FCA平均耗时的0.7%,极大提高了大规模二进制程序模化的效率。Currently modularization of large-scale binary programs costs a lot of time,and the"modules"of software system have certain characteristics similar to the"community structure"of complex network.This paper studies the use of community detection algorithm for modularization of binary programs.At present,there are a lot of community detection algorithms,so a community detection algorithm selection framework is proposed,which includes three selection criteria and four evaluation criteria.According to the selection criteria,three representative community detection algorithms are selected,and the modularization effect of the algorithms are evaluated by the evaluation criteria.Through several groups of comparative experiments,a good community detection algorithm Louvain is selected for modularization of large-scale binary programs.Experimental results show that Louvain is superior to the current best modularization method FCA,and the average experimental time is only O.7%that of FCA,which greatly improves the efficiency of large-scale binary program modularization.

关 键 词：软件模块化 大规模 二进制程序 复杂网络 社团检测 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]