基于源码解析的软件依赖网络构建研究  

Research on Software Dependency Network ConstructionBased on Source Code Parsing

作　　者：刘辉辉 马乐军 王蓁蓁[1] 闫瑞 杨沐雨 LIU Hui-hui;MA Le-jun;WANG Zhen-zhen;YAN Rui;YANG Mu-yu(Jinling Institute of Technology,Nanjing 211169,China;Nanjing University,Nanjing 210093,China)

机构地区：[1]金陵科技学院软件工程学院,江苏南京211169 [2]南京大学软件新技术国家重点实验室,江苏南京210093 [3]金陵科技学院网络安全学院,江苏南京211169 [4]南京大学生命科学学院,江苏南京210093

出　　处：《金陵科技学院学报》2023年第4期1-8,70,共9页Journal of Jinling Institute of Technology

基　　金：金陵科技学院高层次人才科研启动基金(jit-b-202007,jit-b-202211);江苏省高等学校基础科学(自然科学)研究重大项目(22KJA520002)。

摘　　要：软件依赖网络是指代码中各个实体因相互依赖而形成的复杂网络结构。当前,大多数的研究使用商业或闭源工具构建依赖网络数据集,工具之间缺乏统一的实体依赖模型,难以进行数据交换、工具迁移和性能对比。为此,采用源码解析的方式提取依赖关系,构建软件依赖网络,并在6个开源项目上,从依赖提取和依赖强度两个方面开展实证分析,构建一个规模不等、可复现、开放、共享的依赖网络数据集,以便研究人员开展基于软件依赖网络的应用研究。Software dependency network is the complex network structure formed by the interdependence between entities in code.Currently,most studies rely on commercial or proprietary closed tools for constructing dependency network datasets,and a unified entity dependency model is lacking among these tools.This absence makes it difficult to exchange data,migrate tools,and compare performance.Therefore,this paper uses source code parsing to extract dependency relationship,construct software dependency networks,and conduct empirical analysis on six open source projects in terms of dependency extraction and dependency strength.It strives to build a dataset of dependency networks that is of varying scales,replicable,open,and shared,in order to facilitate researchers in conducting application researches based on software dependency networks.

关 键 词：软件依赖网络 源代码解析 代码依赖提取 静态分析 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]