Abstract
Due to its natural incompleteness, a fault hypothesis generator is always considered to produce inconsistent fault hypotheses. In this paper, some notions on testing generated fault hypotheses are described. In order to discard inconsistent fault hypotheses, a technique classifying types of them and pruning them based on the types is proposed. The technique divides fault hypotheses into three types: pseudo fault hypotheses, contradicting fault hypotheses, and candidate faults. Experiments on testing fault hypotheses by pruning the pseudo and contradicting fault hypotheses have been done, and the results have shown that by doing so, inconsistent fault hypotheses can be gotten rid of considerably. Furthermore, notions to deal with candidate faults are presented as well.
Original language | English |
---|---|
Pages | 559-566 |
Number of pages | 8 |
Publication status | Published - 1994 Dec 1 |
Externally published | Yes |
Event | Proceedings of the 9th International Conference on Applications of Artificial Intelligence in Engineering - University Park, PA, USA Duration: 1994 Jul 19 → 1994 Jul 21 |
Other
Other | Proceedings of the 9th International Conference on Applications of Artificial Intelligence in Engineering |
---|---|
City | University Park, PA, USA |
Period | 94/7/19 → 94/7/21 |
ASJC Scopus subject areas
- Software