Mining Control Flow Abnormality For Sofware Bug Isolation

Jiawei Han, Professor of Computer Science, UIUC

Analyzing the executions of a buggy program is essentially a data mining process: Tracing the data generated during program executions may disclose important patterns and outliers that could eventually reveal the location of software bugs. In this paper, we investigate logic bugs, a kind of bugs that rarely incur memory access abnormality but generate incorrect outputs. We show that through mining control flow data generated in software program execution, we could locate logic bugs in many cases without prior knowledge of program semantics. In order to detect the control abnormality, we propose a statistical model that compares the probability distribution of boolean features between correct and incorrect executions. Our approach has achieved encouraging success on a set of testing programs.

Slides