Which crashes should i fix first?: Predicting top crashes at an early stage to prioritize debugging efforts

Dongsun Kim; Xinming Wang; Sunghun Kim; Andreas Zeller; S. C. Cheung; Sooyong Park

doi:10.1109/TSE.2011.20

Which crashes should i fix first? Predicting top crashes at an early stage to prioritize debugging efforts

Dongsun Kim, Xinming Wang, Sunghun Kim, Andreas Zeller, S. C. Cheung, Sooyong Park

Research output: Contribution to journal › Article › peer-review

68 Scopus citations

Abstract

Many popular software systems automatically report failures back to the vendors, allowing developers to focus on the most pressing problems. However, it takes a certain period of time to assess which failures occur most frequently. In an empirical investigation of the Firefox and Thunderbird crash report databases, we found that only 10 to 20 crashes account for the large majority of crash reports; predicting these "top crashes" thus could dramatically increase software quality. By training a machine learner on the features of top crashes of past releases, we can effectively predict the top crashes well before a new release. This allows for quick resolution of the most important crashes, leading to improved user experience and better allocation of maintenance efforts.

Original language	English
Pages (from-to)	430-447
Number of pages	18
Journal	IEEE Transactions on Software Engineering
Volume	37
Issue number	3
DOIs	https://doi.org/10.1109/TSE.2011.20
State	Published - 2011

Keywords

Crash reports
Data mining
Machine learning
Social network analysis
Top crash

Access to Document

10.1109/TSE.2011.20

Cite this

@article{94818eb8d2de4c88a43cad4b8bf4b653,

title = "Which crashes should i fix first?: Predicting top crashes at an early stage to prioritize debugging efforts",

abstract = "Many popular software systems automatically report failures back to the vendors, allowing developers to focus on the most pressing problems. However, it takes a certain period of time to assess which failures occur most frequently. In an empirical investigation of the Firefox and Thunderbird crash report databases, we found that only 10 to 20 crashes account for the large majority of crash reports; predicting these {"}top crashes{"} thus could dramatically increase software quality. By training a machine learner on the features of top crashes of past releases, we can effectively predict the top crashes well before a new release. This allows for quick resolution of the most important crashes, leading to improved user experience and better allocation of maintenance efforts.",

keywords = "Crash reports, Data mining, Machine learning, Social network analysis, Top crash",

author = "Dongsun Kim and Xinming Wang and Sunghun Kim and Andreas Zeller and Cheung, {S. C.} and Sooyong Park",

year = "2011",

doi = "10.1109/TSE.2011.20",

language = "English",

volume = "37",

pages = "430--447",

journal = "IEEE Transactions on Software Engineering",

issn = "0098-5589",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "3",

}

TY - JOUR

T1 - Which crashes should i fix first?

T2 - Predicting top crashes at an early stage to prioritize debugging efforts

AU - Kim, Dongsun

AU - Wang, Xinming

AU - Kim, Sunghun

AU - Zeller, Andreas

AU - Cheung, S. C.

AU - Park, Sooyong

PY - 2011

Y1 - 2011

N2 - Many popular software systems automatically report failures back to the vendors, allowing developers to focus on the most pressing problems. However, it takes a certain period of time to assess which failures occur most frequently. In an empirical investigation of the Firefox and Thunderbird crash report databases, we found that only 10 to 20 crashes account for the large majority of crash reports; predicting these "top crashes" thus could dramatically increase software quality. By training a machine learner on the features of top crashes of past releases, we can effectively predict the top crashes well before a new release. This allows for quick resolution of the most important crashes, leading to improved user experience and better allocation of maintenance efforts.

AB - Many popular software systems automatically report failures back to the vendors, allowing developers to focus on the most pressing problems. However, it takes a certain period of time to assess which failures occur most frequently. In an empirical investigation of the Firefox and Thunderbird crash report databases, we found that only 10 to 20 crashes account for the large majority of crash reports; predicting these "top crashes" thus could dramatically increase software quality. By training a machine learner on the features of top crashes of past releases, we can effectively predict the top crashes well before a new release. This allows for quick resolution of the most important crashes, leading to improved user experience and better allocation of maintenance efforts.

KW - Crash reports

KW - Data mining

KW - Machine learning

KW - Social network analysis

KW - Top crash

UR - http://www.scopus.com/inward/record.url?scp=79957873703&partnerID=8YFLogxK

U2 - 10.1109/TSE.2011.20

DO - 10.1109/TSE.2011.20

M3 - Article

AN - SCOPUS:79957873703

SN - 0098-5589

VL - 37

SP - 430

EP - 447

JO - IEEE Transactions on Software Engineering

JF - IEEE Transactions on Software Engineering

IS - 3

ER -

Which crashes should i fix first? Predicting top crashes at an early stage to prioritize debugging efforts

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this