Premium
A new disjunctive literal insertion fault detection strategy in boolean specifications
Author(s) -
Paul T. K.,
Chowdhury M. J. M.,
Lau M. F.
Publication year - 2021
Publication title -
journal of software: evolution and process
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.371
H-Index - 29
eISSN - 2047-7481
pISSN - 2047-7473
DOI - 10.1002/smr.2336
Subject(s) - literal (mathematical logic) , computer science , expression (computer science) , fault (geology) , boolean function , fault detection and isolation , test suite , algorithm , boolean expression , fault coverage , syntax , test case , reliability engineering , programming language , artificial intelligence , machine learning , regression analysis , electrical engineering , engineering , seismology , electronic circuit , actuator , geology
Abstract In fault‐based Boolean expression testing, the main challenge is to generate effective test cases that can detect faults within expressions. Previous studies show that it is hard to detect literal insertion faults, more specifically Disjunctive literal insertion fault (LIF[ + ]) compared with other faults in Boolean expressions. Researchers have been using different strategies such as multiple near false point coverage (MNFP) and modified condition decision coverage (MCDC) to detect LIF[ + ] faults. However, these strategies have their own limitations. For example, MNFP can only be applied when the expression is in irredundant disjunctive normal form (IDNF), and MCDC detects a low percentage of LIF[ + ] faults. In this paper, we propose an abstract syntax tree (AST)‐based test case generation strategy for LIF[ + ] fault detection that overcomes these limitations. Furthermore, our experimental results indicate that, on average, the test suites satisfying the proposed strategy can detect approximately 97.3% of LIF[ + ] faults for general form expressions and 89.7% of LIF[ + ] faults for IDNF expressions, which are 15.6% and 13.8% improvement, respectively, compared to the MCDC test suites. Moreover, the size of the required test suite is smaller than that of MCDC test suite.