원문정보
초록
영어
In Object-oriented paradigm, classes are the basic units to organize the small applications. The concept of ‘package’ introduced when programmer found it difficult to organize the large application using classes. A good organization of classes into identifiable and collaborating packages eases the understanding, maintenance, testing and evolution of the software. In order to assess and improve the quality of an application during development process, developers and designers use various software metrics. Robert C. Martin presented a set of software metrics to identify poorly designed packages on the basis of certain features like stability, abstraction and coupling of the packages. According to Robert C. Martin, the Packages which are highly interdependent tend to be rigid, not reusable and hard to maintain. These metrics have been used to predict the quality of the packages early in the design phase and hence helps the programmers to reduce the overall cost of development. But these Software metrics neither give the exact value nor defines the sharp boundaries to categorize the packages. In this research work, a decision making system based on Fuzzy inference mechanism is used to categorize the good and bad packages that depends on various factors which are vague in nature. The proposed model is finally validated using the Distance from the Main Sequence ‘D’ metric intended by Robert Martin.
목차
1. Introduction
2. Characteristics of Packages
2.1. Stability
2.2. Abstractness
2.3. Coupling
3. Robert C. Martin’s Metrics
3.1. Afferent Coupling (CA)
3.2. Efferent Coupling (CE)
3.3. Instability (I)
3.4. Abstractness (A)
3.5. Distance from Main Sequence (D)
4. Fuzzy Approach to Categorize the Packages
4.1. Proposed Model
4.2. Working of the Model
4.3. Membership Function for Inputs and Outputs
4.4. Robert Martin’s Rule Base
5. Results of Proposed Model
5.1. Evaluation Metric and Quantitative Results
5.2. Quantitative Results of the Proposed Model
6. Conclusion
References