The core
contribution of our research is a scenario based assessment technique
called SALUTA. (Scenario based Architecture Level UsabiliTy
Assessment technique).
In Deliverable 3.2 we
developed a method for architecture analysis of usability. This method is
based on scenario based assessment i.e. in order to assess a particular
architecture, a set of scenarios is developed that concretizes the actual
meaning of a requirement [Bosch, 2000]. For that purpose usage scenarios are defined. By analyzing the architecture for
its support of each of these usage scenarios we determine the
architecture’s support for usability. SALUTA consistes on the following five steps:
1. Create usage profile; describe
required usability (this step has been described in Deliverable 3.1)
2. Analyze the software
architecture: describe provided usability. (this step has been described in
Deliverable 3.2)
3.
Scenario
evaluation: determine the architecture's support for the usage scenarios. (this step has been described in Deliverable
3.1)
4.
Interpret
the results: draw conclusions from the analysis results. (this step has been described in Deliverable
3.1)
A
detailed description of each step can be found in Deliverable 3.1 and Deliverable 3.2 but
basically SALUTA works as follows: first the usability requirements are
rewritten in the form of a set of scenarios. We then analyze the
architecture to identify patterns and properties that have been
implemented. Then the architectures support for the scenarios is analyzed
based on the evidence of patterns and properties that have been
implemented. The analyst used the SAU framework to reason about how a
particular pattern or property improves or impairs a certain aspect of
usability. Finally either a qualitative or a quantitative result is
formulated based on the assessment results.
In
order to validate SALUTA we have applied this technique at three case
studies. The goal of the case studies was to conduct a software
architecture analysis of usability on each of the three systems. As a
research method we used action research [Argyris et al,
1985], i.e. we took upon our self the
role of external analysts and actively participated in the analysis process
and reflected on the process and the results. One case study has been
performed at a software organization which is part of our university; the
other two case studies are performed at our industrial partners in the
project. These case studies, have been published as part of the STATUS
deliverables in [Folmer et al-a,
2004] and in a pending article [Folmer et al-b,
2004].
Our
cases studies show that it is possible to use SALUTA to assess software
architectures for their support of usability, whether we have accurately
predicted the architecture support for usability can only be answered after
the results of this analysis are compared to the results of final user testing
results when the system has been finished. Validation is difficult for this
assessment technique as many factors are involved in the success or failure
of a software project and it is not possible to do a controlled experiment
to validate the assessment technique.
In
general, our assessment approach was overall well received by the software
architects. In some cases the software architect had not considered the use
of some of our patterns or properties but they considered to implement them
in the system based on the result of the assessment. Even when the usage
profile that was based on the current usability requirement was supported,
architects still considered some of the patterns to be implemented because
of future usability requirements
|