Loading...

Extracting Bugs from Mobile Apps Users’ Reviews and Localizing Bug Referenced in Source Code Files

Ghanadi, Fatemeh | 2020

423 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52641 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Heydarnoori, Abbas
  7. Abstract:
  8. In recent years, with the rise of the smartphones, mobile apps have seen unprecedented growth and their stores have gained popularity among users. The direct connection of users to these stores and the ability to comment on the applications, have made them big repositories of information. In particular, user reviews provide developers with valuable information on strengths and weaknesses of application and users' problems when using it, which can serve as a guide to improve their application.There are a lot of research on exploring user reviews but few have been trying to find a link between reviews and source code of application. In this research, for the first time, all of the code structure information, commits, and application release notes were used to localize reviews containing bug report in source code files. Using the TFIDF model and the cosine similarity between the expressions, the proposed solution searches for the location of the bug in versions, commits, and source code files. Finally, it presents a list of suspicious files by fusing the obtained data. The proposed method was implemented on nine open source applications from Google Play that achieved an average accuracy of 65% in the top five files, 81% in the top 10 files, 0.412 of mean average precision(MAP) and 0.481 of mean reciprocal rank(MRR)
  9. Keywords:
  10. Opinion Mining ; Software Maintenance ; User Reviews ; Hybrid Mobile Applications ; Bug Localization

 Digital Object List

 Bookmark

No TOC