The Research Of Metrics In Object-oriented Sotfware Complexity Based On Class Diagram

Since the20th century, computer brought far-reaching impact on human society.With the rapid development of computer science and information technology,computers are used in more and more areas, more and more data and information isprocessed by computers, the data processing capability of computer is becominghigher and higher. Software developed by computer technology increasing scale, thecomplexity of software is increasing, it is very important and necessary to measure thecomplexity of software effectively and then improve quality of the software. In recentyears, with the continuous development of object-oriented technology, the demand ofsoftware reusability becoming more and more obvious, the key issues is how tomeasure the complexity of object-oriented software effectively. UML becoming theindustry standard in object-oriented modeling field, As the most important UMLmodels, the complexity of class diagram can give a direct impact on the thecomplexity of the entire object-oriented software, so it necessary to find a methodwhich is widely accepted and effective to measure the complexity of class diagram.Firstly, this paper considered the various factors that affect the complexity ofUML class diagrams, use AHP subdivide from the various factors from semanticcategory and then got the complex hierarchy model of class diagram. Secondly,constructed the corresponding relationship between the instance of the class diagramand code according to the code generation rules of the class diagram, defined theweight of various factors through a lot of class diagram instance and code from codeareas,and then got a weight value which consistent with peoples’ practical experience.At last use fuzzy comprehensive evaluation method to measure the complexity ofclass diagram from semantic category and code areas respectively, compare with theresults of the previous measure, the experimental results show that the method of thisarticle is closer to people’s practical experience results, verify the effectiveness of theproposed method.The object of this paper is the class diagram in UML model, integrated use of theanalytic hierarchy process, fuzzy comprehensive evaluation method, mathematicalmodeling theory, uncertain multiple attribute decision theory, measure the complexityof class diagrams through system modeling, fuzzy evaluation.