Abstract


Nowadays, software size is used in the management and control of software production, and is employed as one of the essential parameters for benchmarking productivity and quality of software projects and delivered products. Although the importance of early size assessment is evident, it is currently only achievable in later phases of the software lifecycle (analysis, design, and implementation). 
Software size can be quantified using several approaches such as code lines and Functional Size Measurement (FSM) methods. An FSM method measures the size of software by quantifying the Functional User Requirements. The most widely used FSM method is Function Point Analysis (FPA). This method was developed to measure Management Information Systems (MIS) and assumes the use of traditional software development methodologies such as structured analysis and design.
 Although IFPUG FPA has achieved increasing popularity in the industry, its lack of applicability to all software types and the rapid evolution of development paradigms have produced many variations of this method.
In order to address these weaknesses, COSMIC-FFP has emerged as the second generation of FSM methods. This method is an international standard approved by ISO (ISO/IEC 19761). However, the generality of COSMIC-FFP requires instantiation through a more specific and systematic procedure in conjunction with a software development method.
A number of proposals that use the COSMIC-FFP method to measure the functional size of future software applications from high-level specifications already exist. However, these proposals lack rigorous and systematic definition; for instance, a common problem is the lack of a clear definition of the various concepts that contribute to software size, particularly at the requirements modelling level. Another weakness is that validation is not covered by these proposals, despite the fact that the ISO/IEC 14143-3 standard provides a framework for verifying whether an FSM method complies with specific performance properties (e.g. repeatability, reproducibility, accuracy, etc.).
This thesis confronts these problems directly by analyzing and adapting COSMIC-FFP in order to measure the functional size of a software application from its corresponding requirements specification in the OO-Method context. OO-Method is an automatic software production method, based on a model-transformation strategy.
Therefore, the objective of this thesis is twofold: 1) to define a procedure for the automatic measurement of the functional size at the earliest stage of the software production process using COSMIC-FFP; and 2) to evaluate the quality of this measurement procedure by looking at its design, application, and the results obtained.   To achieve these objectives, a measurement procedure called RmFFP has been systematically defined. To do this we used the process model for software measurement suggested by Abran and Jacquet. 
The thesis dealt with the following: 1) the definition of a set of mapping rules that allows the significant primitives of the Requirements Model to be identified; 2) the definition of a set of measurement rules for obtaining the functional size of OO systems generated using the OO-Method approach; 3) the application of both rules sets to specific real case studies; 4) design validation by means of conformity evaluation with ISO/IEC 19761, theoretical validation using a formal framework, and an analysis carried out from a metrology perspective; 5) evaluation of the results in terms of accuracy and reproducibility; 6) finally, the empirical evaluation of the perceptions of users employing RmFFP to measure early-stage functional size. We have corroborated that RmFFP is perceived as easy to use and useful and that, furthermore, there is an intention to use it in the future.