Model-Based Measurement of Software Layers Using the COSMIC Method.

The management of software development projects relies on software measures. In particular, the measures of the functional size of software are important for estimating the amount of resources needed for development. Several methods have been proposed to measure the functional size of software. These methods aim at measuring functional user requirements, i.e., the functionality of software as perceived by the end user. The COSMIC consortium defined a standard process to perform functional size measurements, and published the official guidelines for COSMIC measurement, which results in a measure expressed in the COSMIC Function Point size unit. Traditionally, a software application is considered as an atomic object, since from the point of view of the end user there is a single entity (the software application) that delivers the required functionality. However, the evolution of software technologies led to software architectures that can be quite complex. At the same time, new methods and notations have been proposed to represent functional user requirements; among these are modelling languages like UML. The COSMIC method is defined so that it is possible to measure either the application as a whole or its structural parts (e.g., the server and client components of a client-server application) separately. It has also been shows that COSMIC measurement can be applied to functional user requirements written as UML diagrams. Not only: measuring well-defined, well-structured requirements written in UML is generally easier and faster than measuring functional requirements written in an unstructured way (e.g., in natural language). However, hardly any attention has been dedicated to the issue of exploiting the architectural information contained in UML models: this information concerns the structure of the software and can be very useful to support the measurement of parts of a software application. In this paper it is shown how to take into account the structural information provided by UML diagrams to perform COSMIC measures concerning parts of an application. The proposed technique is applied to a realistic example proposed by COSMIC as part of their measuring guidelines. [ABSTRACT FROM AUTHOR]