Title :
Orphan Defects: Chance Finding?
Author :
Dhanyamraju S. U. M. Prasad;Satya Sai Prakash Kanakadandi
Author_Institution :
HCL Technol., Noida, India
Abstract :
At least 5% and at most 25% defects escape the rigorous to routine testing practices of various software products. Defects imply various holes in the test suite creation as well as selection. We implemented hybrid distance based algorithms to derive defect to test case mapping that yielded two metrics - Test Coverage Metric and Test Sufficiency Metric. In addition to these two metrics, we stumbled upon a set of defects that are not mapped to any test cases which were named as ´orphan defects (OD)´. Finding orphan defects contrasts the testing team´s belief that they have created a comprehensive test suite with total coverage, we have coined the term ´chance finding´ to empaahsize that it is serendipitous finding while analyzing various defects, both internal and custoemr reported. These defects have been further analyzed for any associated features and presented to the user for validation and adding required test cases. Upon using this procedure, one of the industry partners has brought down the orphan defects by about 90% over three patch releases and test coverage and sufficiency have been improved considerably.
Keywords :
"Feature extraction","Testing","Measurement","Software","Computer bugs","Training","Natural language processing"
Conference_Titel :
Artificial Intelligence with Applications in Engineering and Technology (ICAIET), 2014 4th International Conference on
DOI :
10.1109/ICAIET.2014.22