A Cognitive Complexity Metric of Software

Abstract

Software complexity metrics can be defied as a quantitative process of arriving a decision with a certain property of the software given. The measurements arrives from the software complexity metric helps to identify the weaknesses of the software development process and also its’ quality as well. Although a software complexity metric adheres only for a certain attribute of the given software, it is difficult to arrive with the term of the complexity of it. Therefore the Cognitive Complexity Metrics are mainly used to measure the complexity of the software which goes through how the human brain will understand the given software or the source code in terms of Cognitive informatics with different aspects as well. Since there is not a standard way of determining the Cognitive Complexity if a software, it is expected to come up with a way of calculating the Cognitive Complexity according to three different aspects. The Architectural aspect identifies the amount of information inside the source code given. The Spatial aspect determines how the information is there in the source code in terms of LOC value. Thus the proposed Cognitive Complexity metric is calculated as the summation of above two aspects. Furthermore the same Cognitive Complexity calculation is carried with an implementation of an automated tool which is implemented on the Windows platform. It can browse the corresponding source code to the application and provides its’ Cognitive calculation without performing the manual computation. Within the automated calculation, some Cognitive complexities are calculated by considering the highest complexity path rather than considering all possible executable paths like in manual computation. The third aspect, the User Understandability determines how different users go through the source code given under three different user groups. A software complexity metric should be validated to ensure the usage by providing a good effectiveness for the software to be measured. The proposed Cognitive Complexity metric is evaluated against nine Weyuker properties resulting all the nine properties satisfaction and also with five properties of Briand’s framework resulting three properties satisfaction. Therefore it concludes the validity of its’ practical usage to measure the Cognitive Complexity of a software. Also the proposed metric is used for both Procedural and Object Oriented codes to measure the Cognitive Complexity. It is found that the Cognitive Complexity of an object oriented code is higher than the procedural code which provides the same functionality.