Sung Ae Kim, Wan Kyoo Choi, Sung Joo Lee

The Transactions of the Korea Information Processing Society (1994 ~ 2000), Vol. 7, No. 11, pp. 3372-3383, Nov. 2000
10.3745/KIPSTE.2000.7.11.3372,   PDF Download:

Abstract

In object-oriented paradigm, highly cohesive classes define only a data factor(instance variable) whose all methods(or operations) refer to, or have the maximum connection among elements in a class. Therefore, an increase in the number of the data factors, and the operations to obstruct the maximum connection among elements in a class are the main factors to decrease the class cohesion. In this paper, we propose a new measure, called CM(Cohesion Measure), which can measure the class cohesion more consistently with our intuition on the basis of factors to decrease it. We use the reference graph in order to find the decreasing factors of the class cohesion, and measure the class cohesion on the basis of analysis of the domain reference graph. We also show that the proposed CM metrics satisfy the properties that the cohesion measures should satisfy, and that it can provide more upgraded measures than the existing measures through comparing it with the measures of existing cohesion.

Statistics
Cite this article