Academic journal article Communications of the IIMA

Automated Software Quality Visualisation Using Fuzzy Logic Techniques

Academic journal article Communications of the IIMA

Automated Software Quality Visualisation Using Fuzzy Logic Techniques

Article excerpt

ABSTRACT

In the past decade there has been a concerted effort by the software industry to improve the quality of its products. This has led to the inception of various techniques with which to control and measure the process involved in software development. Methods like the Capability Maturity Model have introduced processes and strategies that require measurement in the form of software metrics.

With the ever increasing number of software metrics being introduced by capability based processes, software development organisations are finding it more difficult to understand and interpret metric scores. This is particularly problematic for senior management and project managers where analysis of the actual data is not feasible.

This paper proposes a method with which to visually represent metric scores so that managers can easily see how their organisation is performing relative to quality goals set for each type of metric. Acting primarily as a proof of concept and prototype, we suggest ways in which real customer needs can be translated into a feasible technical solution.

The solution itself visualises metric scores in the form of a tree structure and utilises Fuzzy Logic techniques, XGMML, Web Services and the .NET Framework. Future work is proposed to extend the system from the prototype stage and to overcome a problem with the masking of poor scores.

INTRODUCTION

With the advent of new methodologies, process-driven management and new project management tools, the number of software metrics in use has increased dramatically (White et al., 2003). Software Process Improvement (SPI) programmes, like the Capability Maturity Model Integration (CMMI), require that metrics are recorded in order to achieve accreditation (Reitzig, et. al, 2003). As organisations strive toward CMMI Level 4 or 5 the number of metrics they must record increases (White et al., 2003). Core to SPI are the metrics produced from measurement of software practices and processes.

This measurement is seen by the Software Engineering Institute (SEI) as pivotal in the improvement of "management and work processes of software development and acquisition" (Goethert and Siviy, 2004). Having implemented these processes, companies often find themselves producing vast quantities of metrics to satisfy the audits they must pass in order to achieve certain levels of compliance with CMMI, but organisations then often fail to make full use of metrics because they struggle to consistently understand indicators derived from measurement data. Their investigation identified numerous challenges faced when organisations attempt to make use of software metrics. The major concerns of the SEI are:

* Information is often misunderstood or misinterpreted

* Lack of data integrity

* No base for comparison

(Goethert and Siviy, 2004)

The problems found are not specific to CMMI, which is a generic approach to achieving and assessing software capability. Rather the problems relate to the capture of metrics across multiple projects within a single organisational domain. The real potential for software metrics is in their analysis so that they can be used to help achieve the goal of any software development through managing risks, and subsequently the process improvement programme by learning to improve the processes used to develop software and hence improve quality and efficiency (Walden, 2002).

Software metrics, therefore, have struggled to penetrate into mainstream software engineering because they were not able to provide management with this kind of information and hence failed to deliver their primary objective (Fenton and Neil, 2000). Indeed, metrics are sometimes viewed as a hindrance to the developers and managers who have to compile them (Alexander, 2003). The future success of metrics lies in bringing together different areas of software development and testing and enables managers to make predictions, assessments and tradeoffs during the project lifecycle. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.