During software development and maintenance stages, programmers have to frequently debug the software. One of the most difficult and complex tasks in the debugging activity is software fault localization. A commonly-used method to fix software fault is computing suspiciousness of program elements according to failed test executions and passed test executions. However, this technique does not give full consideration to dependences between program elements, thus its capacity for efficient fault localization is limited. Our research intends to introduce program slicing technique and statistical method which extracts dependencies between program elements and refines execution history, then builds program slicing spectra to rank suspicious elements by a statistical metric. We expect that our method will contribute directly to the improvement of the effectiveness and the accuracy of software fault localization and reduce the software development and maintenance effort and cost.
展开▼
