Detail

Developing Software Technologies through Experimentation: Experiences from the Battlefield

For a long time scientists have been committed to describe and
organize information obtained by observations from the field. The ideas
regarding experimentation have been explored and evolved into different
scientific areas, including physics, agriculture, medicine, engineering and
social sciences among others. We can observe similar behavior in Software
Engineering. Software engineers have intensively worked to understand the
application and evolution of software processes and technologies by applying
the scientific method (experimentation) to support their researches.
Nowadays, experimentation has demonstrated to be a need for the evolution of
the field, representing a cardinal tool to allow the transference of
software technologies to the industry; improve software processes and
evidence behaviors in Software Engineering. The Experimental Software
Engineering Group at COPPE/UFRJ (ese.cos.ufrj.br) promotes experimentation
in the building of software technologies to the industry.

       To make a software technology available to the industry demands three
stages: its conception, construction and evaluation. Some approaches to
construct and evaluate software technologies are easily found in the
technical literature. However, this is not the case with the conception
stage. This talk discuss an approach concerned with the conception,
construction and evaluation of software technologies based on evidence
collected through secondary and primary studies used by ESE group. The types
of study (including quasi systematic reviews, surveys, quasi-experiments,
action research and so on) and decision points are presented to allow its
evaluation and application by software engineers. The discussion will be
illustrated with concrete examples of technologies that have been developed
following this approach.