Heather M. Conboy

Postdoctoral Researcher

College of Information and Computer Sciences

University of Massachusetts Amherst

Curriculum Vitae


Contact Information

140 Governors Drive
Amherst, MA 01003-9264
USA

Email: hconboy@cs.umass.edu
Office: 316 Computer Science Building
Telephone: 1-413-545-2146


Research

I work in the Laboratory for Software Engineering Research (LASER) with my co-advisors George Avrunin and Lori Clarke. My research interests are in the areas of requirements engineering, model checking, and analysis.

The goal of my PhD research has been to investigate techniques to improve the quality of human-intensive processes (HIPs) that involve coordination among human performers, software applications, and hardware devices. For instance, consider an in-patient surgery medical process that involves a surgeon, anestheologist, nurses, patient, electronic order entry system, and an infusion pump. Since the HIPs are often complex involving aspects such as concurrency and exceptional situations, I have developed static and dynamic analysis techniques that help prevent the HIPs from violating their overall system requirements and evaluated these techniques on either benchmarks or case studies.

For my masters work, I contributed to the FLAVERS model checker that employs static data flow analysis techniques to verify whether or not all potential traces through a finite-state model of a system satisfy a given requirement of that system. If not, this model checker can generate a counterexample trace that illustrates a potential system requirement violation to help with program understanding and debugging. For my thesis work, I am investigating a static analysis approach to automatically derived requirements for components used in HIPs based on interface synthesis methods developed for software systems. The derived requirements are represented as finite state automata that prevent any sequences of procedure calls to a selected component that violate a given system requirement. Such requirements can be used to select the right process for a given component or alternatively to select the right component for a given process. Since checklists that guide the human performers though on-going processes have been applied in multiple domains and shown to reduce human errors, I am also contributing to a dynamic approach that generates context-sensitive Smart Checklists by matching validated HIP models against streams of real-world process activity events.

Publications

All of my publications are available from the LASER publication database.

Refereed Papers

Invited Talks and Posters

Software Tools and Benchmarks