期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2016
卷号：5
期号：11
页码：19198
DOI：10.15680/IJIRSET.2016.0511089
出版社：S&S Publications
摘要：High cohesion for classes is one of the most desire properties in Object Oriented (OO) analysis as itsupports program comprehension, testing, maintainability, reusability. Cohesion metrics have been used for qualityvaluation, software modularization, and fault prediction and so on. Existing approaches of class cohesion metrics arelargely based on the structural information of the source code, like attribute references in class methods. Only lookingat particular aspect of cohesion is not sufficient for completely and accurately specifying class cohesion and there is aneed to other aspects of cohesion like conceptual aspect. The two conceptual cohesion metrics ACSM+ and LCSM2#use more efficient document modeling technique, known as LDA, used for defining similarity between methods.Maximum Weighted Entropy (MWE) has two issues, first does not take into account prominent object orientedconcepts like inheritance and second MWE does not make difference between access methods and other methodstereotypes which can artificially increase or decrease cohesion. Extended the MWE by considering the concept ofinheritance and the effect of getter and setter methods and called it MWE+. Two case studies on an open source javasoftware system called Rhino are performed. First case study specifies that the new cohesion metrics capture differentaspects of class cohesion compared to the exiting cohesion metrics. Second case study specifies new dimensions thatMWE+ adds to MWE.
关键词：Metric; Class Cohesion; Topic Modeling; LDA; MWE.
