Research Seminar - August 06, 2004

An Investigation into the value of usage measurement in Empirical Software Engineering Experiments

Time: 11am Friday August 6

Computer Science & Software Engineering

Seminar Room 1.24

Speaker:
A/Prof Richard Thomas

Abstract:

The field of empirical software engineering encompasses both field studies and controlled experiments that address the relationship between entities in software engineering. Thus for example field studies may use CVS repositories as a source of data to test hypotheses about software process. In this talk I shall report on a controlled experiment to discover the effects of various Think Aloud protocols on people undertaking software maintenance tasks.

One of the supplementary goals of the experiment was to discover whether generic usage measurement could improve the data analysis process. Rapid benefits in validation of experimental procedure and the measurement of solution times were observed. More expensively it seems that solution strategies may be partially obtainable as well. Finally some promising results on low level keying rates were obtained.