Better Code Search and Reuse for Better Program Repair

Created by W.Langdon from gp-bibliography.bib Revision:1.8051

@InProceedings{Xin:2019:GI,

• author = "Qi Xin and Steven Reiss",
• title = "Better Code Search and Reuse for Better Program Repair",
• booktitle = "GI-2019, ICSE workshops proceedings",
• year = "2019",
• editor = "Justyna Petke and Shin Hwei Tan and William B. Langdon and Westley Weimer",
• pages = "10--17",
• address = "Montreal",
• month = "28 " # may,
• publisher = "IEEE",
• keywords = "genetic algorithms, genetic programming, genetic improvement, automated program repair, APR, code search, code reuse",
• isbn13 = "978-1-7281-2268-7",
• URL = "http://geneticimprovementofsoftware.com/paper_pdfs/xin2019better.pdf",
• URL = "https://dl.acm.org/citation.cfm?id=3339020.3339023",
• DOI = "doi:10.1109/GI.2019.00012",
• acmid = "3339023",
• size = "8 pages",
• abstract = "A branch of automated program repair (APR) techniques look at finding and reusing existing code for bug repair. ssFix is one of such techniques that is syntactic search-based: it searches a code database for code fragments that are syntactically similar to the bug context and reuses such code fragments to produce patches. The keys to its success lie in the approaches it uses for code search and code reuse. We investigated the effectiveness of ssFix using the Defects4J bug dataset and found that its code search and code reuse approaches are not truly effective and can be significantly improved. Motivated by the investigation, we developed a new repair technique sharpFix that follows ssFix's basic idea but differs significantly in the approaches used for code search and code reuse. We compared sharpFix and ssFix on the Defects4J dataset and confirmed through experiments that (1) sharpFix's code search and code reuse approaches are better than ssFix's approaches and (2) sharpFix can do better repair. sharpFix successfully repaired a total of 36 Defects4J bugs and outperformed many existing repair techniques in repairing more bugs. We also compared sharpFix, ssFix, and four other techniques on another dataset Bugs.jar-ELIXIR. Our results show that sharpFix did better than others and repaired the largest number of bugs.",
• notes = "ssFix, source code matching from repository, code reuse. SharpFix. 103 examples from Defects4j M69. GZoltar spectrum-based fault localisation. Code search NLP n-grams TF IDF. Code translation (name identifier fix up) and transformations.

  Georgia Institute of Technology

  Slides: http://geneticimprovementofsoftware.com/slides/xin2019better_slides.pdf

  GI-2019 http://geneticimprovementofsoftware.com

  part of \cite{Petke:2019:ICSEworkshop}",

}

Genetic Programming entries for Qi Xin Steven Reiss

Citations