Dr. James D. Arthur

Associate Professor
of
Computer Science

(540) 231-7538
arthur@vt.edu
----------

Education:

Current Research Interests:

Recent Research Grants and Contracts:

Professional Activities:

Awards and Honors:

Other Professional Experience:

Selected Publications:

My research interests lie in two domains: software engineering and parallel computation.

Within the software engineering domain current research efforts proceed on three complementary fronts - software quality assessment, independent verification and validation, and requirements engineering.

In the software quality assessment area the focus is on systematic quality assessment guided by the Objectives/Principles/Attributes (OPA) Framework for software development. Software Quality Indicators, undeniably linked to software quality attributes, form a basis for measurement. Substantiated links between desirable software engineering objectives and the principles employed to achieve those objectives, and product attributes resulting from a process guided by such principles, enable the propagation of indicator values leading to quality assessment at multiple levels. The OPA Framework also supports a rational approach for reasoning about how and why particular measures are observed.

Verification and validation (V&V) techniques can provide significant benefits and insights when used in tandem with software development activities. Unfortunately, powerful V&V techniques are often ignored because of the time and costs associated with their use. Our research questions "why" the prohibitive cost and time, and "how" to reduce the cost and time without significantly sacrificing benefits. Of particular interests is developing V&V techniques that support the evolution of requirements that reflect what the user intends to convey as opposed to what is often (mis)understood by the person developing the requirements.

The work in IV&V has brought to light the real importance of acquiring the correct set of requirements from the customer, that is, requirements that reflect the intent of the customer. Initial work has revealed two major factors contributing to our inability to elicit requirements meeting the customer's intent: the lack of structure and lack of guidance in the requirements generation process. To that end, we have defined a Requirements Generation Model (RGM) that consistes of a multiphase framework, a comprehensive set of guidelines and protocols that structure interaction, and a monitoring methodology that continuously "samples" the interaction for deviations from the expected norm and suggests corrective actions when such deviations are detected. We are currently in the process of validation the RGM and are moving toward developing an environment supporting the RGM.

My interests in parallel computation stem from the desire to provide an intuitive approach to specifying parallel computation without being encumbered by unnecessary language complexities or architectural details. In support of this interest, research efforts are being directed toward optimization techniques for computational systems based on the Linda coordination language. More specifically, the optimizations being examined are intended to be transparent to the user, and focused on the exploitation of program composition, language semantics and network architectures.

----------

Personal Interests