Software Design Consistency Checking of UML Models using Artificial Intelligence Techniques

Abstract

Modern human life is made convenient, connected and controlled newlineby software. Software industry is the biggest industry in terms of newlinemanpower and turnover that requires true engineering skills. Software newlineindustry has a profound influence on economy. Software forms the core newlineof manufacturing, business, education, travel, healthcare and many more. newlineThe software developers are faced with the challenge of developing newlinequality systems. newlineGood designs are the key to developing successful softwares. newlineSoftware design can be systematically represented with Unified Modeling newlineLanguage (UML). About 89% of the practitioners use UML for modeling newlinesoftware artifacts, 55% uses UML to design system or code and 77% use newlineUML or graphical notations for software maintenance. The design newlinecomprises of different interrelated design diagrams expressing the newlinedynamic and static aspects of the software. Inconsistency in UML models newlineoccurs when two or more diagrams describe different aspects of the newlinesystem and they are not jointly satisfiable. Manual detection of newlineinconsistencies in large complex systems are incomplete and error prone. newlineArtificial intelligence techniques can replace the manual efforts to make newlinethe development of software easier and cost effective. In this thesis, we newlineclearly demonstrate how intra-model inconsistencies among UML class, newlineactivity, sequence and state diagrams are handled. We introduce a novel newlinemethod of modeling inconsistency handling as an optimization problem. newlineWe have defined maximizing fitness functions that compute the fitness newlinevalues of the overlapping model elements. The fitness value is computed newlineas a function of the properties of the overlapping model elements. newlineInconsistencies are identified from the fitness values. The inconsistencies are newlinefixed during iterations of the self regulating particle swarm optimization newline(SRPSO) algorithm. newline