Welcome to the AWARE Research Website!

This website provides an overview of the Automated Warning Application for Reliability Engineering (AWARE) research at North Carolina State University. This research is being conducted by Sarah S. Heckman, Lucas Layman, and Stephen Thomas under the advising of Dr. Laurie Williams. Funding for this research is provided by the Center for Advanced Computing and Communication and an IBM PhD Fellowship awarded to Sarah for the 2006-2007 and 2007-2008 academic years.

Overview

Automated static analysis is useful for finding common programming mistakes that may lead to field failures. However, automated static analysis generates a large number of false positive alerts, which are alerts that are not potential faults in the source code. Therefore, a manual inspection of each static analysis alert (a notification to the developer, in the form of a warning message, of a potential problem, or fault, in the source code that has been identified via static analysis) is required to determine if the alert is a fault in the software. A large number of false positive alerts discourages the use of static analysis by developers.

The objective of Sarah's research is to create and validate an adaptive ranking model to rank alerts generated by automated static analysis to maximize the number of true positive alerts that appear at the top of an alert listing. The Adaptive Ranking Model (ARM) leverages historical feedback from the developer to predict the likelihood an alert generated from automated static analysis is a true or false positive.

AWARE aggregates automated static analysis alerts and ranks the alerts by the likelihood an alert is a true or false positive using the ARM. The alerts are ranked by the likelihood an alert is a true or false positive in the system. The alert ranking is adapted by developer feedback in the form of suppressing alerts found to be a false positives and closure of alerts found to be true positives.