Design Methodology of Networked Software Evolution Growth Based on Software Patterns.
Keqing He,Rong Peng,Jing Liu,Fei He,Peng Liang,Bing Li
DOI: https://doi.org/10.1007/s11424-006-0157-6
2006-01-01
Abstract:Recently,some new characteristics of complex networks attract the attentions of scientistsin different fields,and lead to many kinds of emerging research directions.So far,most of the researchwork has been limited in discovery of complex network characteristics by structure analysis in large-scalesoftware systems.This paper presents the theoretical basis,design method,algorithms and experiment results ofthe research.It firstly emphasizes the significance of design method of evolution growth for networktopology of Object Oriented（OO）software systems,and argues that.the selection and modulationof network models with various topology characteristics will bring un-ignorable effect on the processof design and implementation of OO software systems.Then we analyze the similar discipline of“negation of negation and compromise”between the evolution of network models with different topologycharacteristics and the development of software modelling methods.According to the analysis of thegrowth features of software patterns,we propose an object-oriented software network evolution growthmethod and its algorithms in succession.In addition,we also propose the parameter systems for OOsoftware system metrics based on complex network theory.Based on these parameter systems,it cananalyze the features of various nodes,links and local-world,modulate the network topology and guidethe software metrics.All these can be helpful to the detailed design,implementation and performanceanalysis.Finally.we focus on the application of the evolution algorithms and demonstrate it by a casestudy.Comparing the results from our early experiments with methodologies in empirical software engi-neering,we believe that the proposed software engineering design method is a computational softwareengineering approach based on complex network theory.We argue that this method should be greatlybeneficial for the design,implementation,modulation and metrics of functionality,structure and per-formance in large-scale OO software complex system.