Using the COSMIC Functional Size Measurement Method (ISO 19761) as a Software Requirements Improvement Mechanism
Thesis presented to École de Technologie Supérieure in partial fulfillment of the requirements for the degree of Doctor of Philosophy
This research project investigates the contribution of the COSMIC software functional size measurement method (ISO 19761) to identify defects in functional requirements. The functional size measurer has to understand and interpret the functional requirements used in input to the measurement process. Industry requirements are typically written in natural language and are prone to ambiguities: therefore, interpretation errors do happen. Errors in functional requirements affect the cost of subsequent steps of the software life cycle through rework when these errors are detected and corrected. Organizations then apply review and inspection techniques to detect errors in requirements. However, these techniques do not detect all defects in a single cycle of review or inspection, leaving a number of residual defects in the requirements document. Notwithstanding the application of such techniques, a measurer using the COSMIC method can identify defects in the functional requirements that do not clearly define the elements required as input to the measurement activity.
The objectives of this research project are to quantify the efficiency and effectiveness (in terms of unit cost) when using the COSMIC method as a technique for identifying defects in these functional specifications, comparing the results with an inspection method. The efficiency and unit costs are computed using the number of defects identified and the effort to do so.
Results show that, on average, a measurer participating in an inspection finds a number of defects similar to the number of defects found by the addition of an inspector, maintaining the efficiency. The average unit cost increases slightly as the effort is slightly higher, except that for this cost, the project gets the software functional size usable for estimation, benchmarking, and process improvement monitoring. Without the identification of defects, the functional size measurement is considered as a management cost. With the additional benefit of the identification of defects, the functional size measurement creates added value in terms of rework costs savings since the defects identified by a measurer can be corrected earlier in the development cycle.
In the experiments with experts and practitioners whose experience was limited, it was also observed that the practitioners newly trained to functional size measurement faced several challenges which impacted the quality of their measurement results. Therefore, this research project also contributed in defining requirements for the training of measurers.