Loading...

A Technique for Program Fault Location Using External Information Sources

Nosrati, Ali | 2017

1744 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 50022 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Mirian Hossein Abadi, Hassan; Habibi, Jafar
  7. Abstract:
  8. In recent years, there have been many studies aimed at finding program defect using information of previous ones. These studies, by using the program text, previous bugs, and information related to version control and project management systems create an defect prediction model to find other defect prone areas in the program. The output of defect prediction methods is part of the program’s text, which is likely to have defects. Considering that the result of this method is just the text of the program and does not provide any extra information for finding defects to the user, these methods have not yet been applied to the software industry. On the other hand, by spreading the use of Q & A sites, these sites have become rich and up-to-date information resources for programmers. In the other words programmers can put their problem along with the code and other information on these sites, so that others can respond for solving them with their own solution. Therefore, in recent years, researchers have been used these information for documenting user interfaces or finding solution for error occurrences in the program. In this thesis, due to problems with the use of defect prediction, a method is presented to help with the information of the question and answer sites, so that the exact location of the defect is identified and, if possible, be presented to the user with the appropriate solution to resolve the defect. In this method, first, using the previous defect information, defect prediction is performed on the program, then the code of the Stack- Overflow question is extracted and using the clone detection methods, the similarities between these code strings and the output of the defect prediction are determined. In the following, in the two stages of the refinement, using the textual connection of the question and the text of the code, as well as the relationship between the text of the code and the answer to the question, the results will be reduced and the final results will be delivered to the user as possible defects. As a result, implementing this method on 5 projects identifies all defects which there is a related question with code segment in StackOverflow
  9. Keywords:
  10. Fault Location ; Information Resources ; Stack Overflow Websites ; Defect Prediction ; Question-Answer Sites

 Digital Object List

 Bookmark

...see more