作者: Jehad Al Dallal
DOI:
关键词: Cohesion (computer science) 、 Empirical research 、 Software quality 、 Data mining 、 Restructuring 、 Information retrieval 、 Class cohesion 、 Object-oriented programming 、 Computer science 、 Java 、 Software
摘要: Classes are the basic units in object-oriented programs, and therefore, their quality has impact on overall of software. Class cohesion is a key factor, it refers to degree relatedness class attributes methods. Software developers use measure assess products guide restructuring poorly designed classes. Several metrics proposed literature, considering special methods (i.e., constructors, destructors, access methods) calculation not empirically studied for most them. In this paper, we address issue one popular metrics, referenced as Lack Cohesion (LCOM). Our empirical study involves applying metric with without classes two open source Java applications statistically analyzing results. The results show that ability LCOM indicating slightly improves when excluding from computation.