Zachary P . Fry – Research Statement

My research focuses on studying the software maintenance process, especially as it relates to humans, in an effort to reduce the overall cost of maintaining large systems. I feel that this problem is of paramount importance as maintenance accounts for the great majority of the cost associated with the software lifecycle. It has remained largely a human-centric, manual process despite increasing levels of automation in related fields. For instance, while techniques have been developed to manage bugs and system understandability, the problem continues to grow as systems become larger and more complex. Thus, I believe more can be done to ease this burden in practice. As such, I have studied the software maintenance process from beginning to end, focusing on finding and fixing bugs while also ensuring continued system understandability and maintainability. My work emphasizes automation wherever possible as well as strong, concrete, humancentric evaluations to better understand both the problems and their solutions. The work I have done to date has two main thrusts: first, I attempt to alleviate known software maintenance bottlenecks by providing automated solutions to difficult, often manual problems. Secondly, I study human-centric concepts related to understandability and continued system quality. For example, finding and fixing bugs is one of the primary concerns associated with software maintenance. As systems grow larger and more complex, the number of bugs often grows proportionally and may become harder to manage. Recognizing the magnitude and difficulty of these problems, I focus on automating parts of the bug localization and bug fixing processes to reduce the human burden associated with these maintenance tasks. Examples of my work in this area include clustering similar defect reports, automatically localizing manually reported bugs, and improvements to an automatic program repair technique. I believe a full examination of the software maintenance process must also explore secondary maintenance concerns including how humans find and fix bugs as well as what aids in making code more understandable. To this end, I have completed user studies examining both fault localization and code understandability. Following these philosophies, my two high-level research goals are as follows: