| In the process of software development, software debugging and maintenance is taking larger percentage of the overhead, in which software debugging is the most difficult and time-consuming task. During the debugging period, programmers need to run the program iteratively and try to find out the possible errors in the program for modifying. As complex diversification and expansion in software, software debugging is becoming more and more complicated, semi-automatic or automatic program fault localization technique arises in time and gradually increases the efficiency of program fault localization. Program fault localization based on the spectrum is one of the popular research hotspots in recent years, which runs some test cases and gets the program running trajectory information, thus calculating the suspiciousness of each program element for locating faults. The gathering process of information on the test coverage is simple, while the locate efficiency is relatively high. However, when it comes to a variety of faults caused by the same one test failure, due to the lack of context information, programmers find it hard to judge the failed test cases with merely one statement, which will cause the overlay of failed test cases and reduce the locate efficiency.This paper introduces software fault localization based on the program spectrum as well as the theory of concept lattice. On the basis of fault localization based on the spectrum, considering the lack of the program fault localization technology based on the application spectrum and concept lattice, entropy in information theory is used in the process of calculating the suspicious degree, thereby creating a new type of suspiciousness measurement model, which expand the measurement of suspiciousness range. In the Eclipse development environment, experiments are conducted in the Print tokens-Java program in Siemens test suites and the Russian game Tetris, which verifies the validity of the proposed method. |
PDF Full Text Request