Investigating the Role of Use Cases in the Construction of Class Diagrams

Author(s): Bente Anda and Dag I.K. Sjøberg
Venue: Empirical Software Engineering
Date: 2005

Type of Experiement: Controlled Experiment
Sample Size: 53
Class/Experience Level: Professional

Quality
3

Summary:
This paper analyses different approaches for building class diagrams from requirements documents. The authors present two experiments to compare use-case driven development with an alternative technique of identifying classes from a textual requirements specification and subsequently apply the use case model to validate the resulting class diagram.

Experiment
The first experiment used 53 professional software engineers. All participants were prompted to construct a class diagram for a library system given a set of requirements documents. Half of the participant were given a professional modeling tool, while the other half were only allowed to use paper and pen. The second experiment used 22 4th year software engineering students. Again the participants were prompted to construct a class diagram for a library system given a set of requirements documents. However all of the participants in this experiment were allow to use the professional modeling tool. The metrics being evaluated in the experiment were: completeness, structure, and time.

Results
Class diagrams constructed with the modeling tool were more complete then class diagrams constructed with pen and paper. However the participants who used the modeling tool also spent more time then the participant who used pen and paper. The experiments “showed that the validation technique resulted in class diagrams that implemented more of the requirements. The derivation technique resulted in class diagrams with a significantly
better structure than did the validation technique in the student experiment and slightly better structure in the experiment with the professionals. There was no difference in time spent between the two techniques.”

0