Scalable statistical bug isolation
Proceedings of the 2005 ACM SIGPLAN conference on Programming language design and implementation
Why Programs Fail: A Guide to Systematic Debugging
Why Programs Fail: A Guide to Systematic Debugging
SOBER: statistical model-based bug localization
Proceedings of the 10th European software engineering conference held jointly with 13th ACM SIGSOFT international symposium on Foundations of software engineering
Empirical evaluation of the tarantula automatic fault-localization technique
Proceedings of the 20th IEEE/ACM international Conference on Automated software engineering
Cooperative bug isolation
Statistical debugging: simultaneous identification of multiple bugs
ICML '06 Proceedings of the 23rd international conference on Machine learning
ICDM '06 Proceedings of the Sixth International Conference on Data Mining
Statistical debugging using compound boolean predicates
Proceedings of the 2007 international symposium on Software testing and analysis
HOLMES: Effective statistical debugging via efficient path profiling
ICSE '09 Proceedings of the 31st International Conference on Software Engineering
Is non-parametric hypothesis testing model robust for statistical fault localization?
Information and Software Technology
Argus: online statistical bug detection
FASE'06 Proceedings of the 9th international conference on Fundamental Approaches to Software Engineering
Hi-index | 0.00 |
Existing Statistical fault localization approaches locate bugs by testing statistical behavior of each predicate and propose fault relevant predicates as nearest points to faults. In this paper, we present a novel statistical approach employing a weighted graph, elicited from run-time information of a program. The predicates are considered as nodes; an edge is denoting a run-time path between two predicates and its label is the number of simultaneous occurrence of connected predicates in the run. Firstly, a typical graph, representing failed run is contrasted with whole graphs of passed runs to find the two most similar graphs of the passed runs and failed runs and discriminative edges are chosen as suspicious edges. In next phase, we statistically test the distribution of the suspicious edges to find the most fault relevant edges; to this end, we apply a normality test on the suspicious edges and based on the test result, we use a parametric or non-parametric hypothesis testing to discover the most fault relevant edges. We conduct the experimental study based on Siemens test suite and the results show the proposing approach is remarkable.