Extracting Significant Specifications from Mining through Mutation Testing (Technical Report)
No Thumbnail Available
Date
2011-07-12T01:42:26Z
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Specification mining techniques are used to automatically infer interaction specifications among objects in the format of call sequences, but many of these specifications can be meaningless or insignificant. As a consequence, when used in program testing or formal verification, the presence of these leads to false positive defects, which in turn demand much effort for manual investigation. We propose a novel process for determining and extracting significant specifications from a set of mined specifications using mutation testing. The resulting specifications can then be used with program verification to detect defects with high accuracy. To our knowledge, this is the first fully automatic approach for extracting significant specifications from mining using program testing. We evaluate our approach through mining significant specifications for the Java API and use them to find real defects in many systems.