CLEAR

From Dependability

Miscommunication of software requirements has been implicated in a disproportionate number of faults and failures in safety-critical and other systems, as well as in schedule and cost overruns that further burden resources. Managing miscommunication is arguably the single most significant problem facing the requirements engineering community; to a first approximation, the greatest improvement to safety-critical software quality and to the efficiency of the development process is to be gained if a substantial advance in requirements accuracy can be achieved.

The CLEAR Method, Cognitive Linguistic Elicitation and Representation, is a methodology under development at the University of Virginia for eliciting and representing software requirements such that the possibility of critical forms of miscommunication is minimized. Its novelty and power derive from its exploitation of research results from linguistics and psychology that allow systematic characterization, identification, elimination and prevention of sources of critical miscommunication. CLEAR directs and supports the collection and management of application-specific domain knowledge critical to the validity of a software requirements specification, via processes and artifacts expressly designed to reduce the potential for breakdown in communication of this domain knowledge by systematically compensating for normal variability in the ways that humans apply communicative heuristics.

Results of quantitative and qualitative assessments indicate that use of the CLEAR Method can be an effective strategy for the collection and dissemination of domain knowledge with higher fidelity, and thus for a reduction in the number of safety-critical defects and process inefficiencies that result.

Selected Papers

• Hanks, Kimberly S., John C. Knight, Elisabeth A. Strunk, and Sean R. Travis

Tools Supporting the Communication of Critical Application Domain Knowledge in High Consequence Systems Development

SAFECOMP 2003, The 22nd International Conference on Computer Safety, Reliability and Security, Edinburgh, Scotland (September 2003) (PDF)

• Hanks, Kimberly S., John C. Knight

Improving Communication of Critical Domain Knowledge in High-Consequence Software Development: an Empirical Study.

21st International System Safety Conference (ISSC'03), Ottawa, Canada (August, 2003) (PDF)

• Hanks, Kimberly S., John C. Knight, and Elisabeth A. Strunk

Erroneous Requirements: A Linguistic Basis for Their Occurrence and an Approach to Their Reduction.

Software Engineering Workshop, NASA Goddard Space Flight Center, Greenbelt, MD (November, 2001) (PDF)