On the functional size measurement of object-oriented conceptual schemas: design and evaluation issues

Functional Size Measurement (FSM) methods are intended to measure the software size by quantifying the functional user requirements. The capability to accurately quantify the software size in an early stage of the development lifecycle is critical for evaluating risks, developing project estimates and having early project indicators. Despite wide spread practitioner acceptance, the FSM methods have been criticized for their inability to correctly measure the size of object-oriented systems. Moreover, most organizations have a number of Web Applications for which functional size must be measured. None of the ISO-standardized FSM methods were designed taking the particular features of Web applications into account. In this thesis, OO-Method Function Points (OOmFP) is presented as a measurement procedure for modeling and sizing object-oriented systems developed by an automatic software production method called OO-Method. This procedure maps the concepts used in OO-Method onto the concepts used by Function Point Analysis, a standard FSM method supported by the International Functional Point Users Group (IFPUG FPA). Additionally, OOmFP is extended to size Web applications. We called this extension OOmFP for the Web (OOmFPWeb), and it measures the functional size of Web applications from conceptual schemas that are specified with Object-Oriented Web Solutions modeling approach (OOWS). We describe the design and the application of the proposed measurement procedure following the steps of a process model for software measurement (Jacquet and Abran, 1997). This thesis also reports on the evaluation of OOmFP and OOmFPWeb. This comprises the validation of the design of the measurement procedures (theoretical validation) and the validation of the application of the measurement procedures (empirical validation). We show the first results towards the theoretical validation of the OOmFP and OOmFPWeb functional size measures using a formal framework called DISTANCE (Poets and Dedene, 1999). With regard to empirical validation, we report the results of two laboratory experiments. In the first experiment, OOmFP is compared to IFPUG FPA using a range of performance-based (i.e., efficiency, reproducibility, accuracy) and perception-based variables (i.e., perceived ease of use, perceived usefulness and intention to use). The goal is to determine whether OOmFP results in better size assessments and is more likely to be adopted in practice than IFPUG FPA. An important contribution is the development and empirical testing of a theoretical model for evaluating FSM methods in general. The results show that OOmFP is more time-consuming than IFPUG FPA, but the measurement results are more reproducible and accurate. In the second experiment, OOmFPWeb is evaluated on a range of variables, including productivity, reproducibility, perceived ease of use, perceived usefulness and intention to use. The results show that OOmFPWeb is efficient when compared to current industry practices. Furthermore, it produces reproducible functional size assessments and is perceived to be easy to use as well as useful in the context of the OOWS development process. Lastly, part of the results of this thesis were put into practice at CARE Technologies S. A. Specifically, the measurement procedure for sizing object-oriented systems was personalized and automated to be used with OlivaNova Model Execution®, a model-based code generation environment.