Extended comprehensive study of association measures for fault localization

摘要

Spectrum-based fault localization is a promising approach to automatically locate root causes of failuresrnquickly. Two well-known spectrum-based fault localization techniques, Tarantula and Ochiai, measurernhow likely a program element is a root cause of failures based on profiles of correct and failed programrnexecutions. These techniques are conceptually similar to association measures that have been proposed inrnstatistics, data mining, and have been utilized to quantify the relationship strength between two variablesrnof interest (e.g., the use of a medicine and the cure rate of a disease). In this paper, we view fault localizationrnas a measurement of the relationship strength between the execution of program elements and programrnfailures. We investigate the effectiveness of 40 association measures from the literature on locating bugs.rnOur empirical evaluations involve single-bug and multiple-bug programs. We find there is no best singlernmeasure for all cases. Klosgen and Ochiai outperform other measures for localizing single-bug programs.rnAlthough localizing multiple-bug programs, Added Value could localize the bugs with on average smallestrnpercentage of inspected code, whereas a number of other measures have similar performance. The accuraciesrnof the measures in localizing multi-bug programs are lower than single-bug programs, which provokesrnfuture research. Copyright © 2013 John Wiley & Sons, Ltd.