by Mouna Hammoudi, Christoph Mayr-Dorn, Atif Mashkoor, Alexander Egyed
Abstract:
Requirement-to-method traces reveal the code location(s) where a requirement is implemented. This is helpful to software engineers when they have to perform tasks such as software maintenance or bug fixing. Indeed, being aware of the method(s) that implement a requirement saves engineers' time, as it pinpoints the exact code region that needs to be edited to perform a bug fix or a maintenance task. Engineers produce traces manually as well as automatically. Nevertheless, traces are incomplete. This limits the amount of information that could be used by an automated technique to check further traces. Therefore, since traces are incomplete, we would like to study the effect of incompleteness on the automated assessment of requirement-to-method traces. In this paper, we apply machine learning on either incomplete or complete tracing information and we evaluate the effect of incompleteness on checking trace information. We demonstrate that the use of complete traces might yield a higher precision but yields a lower recall. Also, the use of incomplete traces yields a higher recall but a lower precision.
Reference:
On the effect of incompleteness to check requirement-to-method traces (Mouna Hammoudi, Christoph Mayr-Dorn, Atif Mashkoor, Alexander Egyed), In SAC '21: The 36th ACM/SIGAPP Symposium on Applied Computing, Virtual Event, Republic of Korea, March 22-26, 2021 (Chih-Cheng Hung, Jiman Hong, Alessio Bechini, Eunjee Song, eds.), ACM, 2021.
Bibtex Entry:
@Conference{DBLP:conf/sac/HammoudiMME21,
author = {Mouna Hammoudi and Christoph Mayr-Dorn and Atif Mashkoor and Alexander Egyed},
booktitle = {SAC '21: The 36th {ACM/SIGAPP} Symposium on Applied Computing, Virtual Event, Republic of Korea, March 22-26, 2021},
title = {On the effect of incompleteness to check requirement-to-method traces},
year = {2021},
editor = {Chih-Cheng Hung and Jiman Hong and Alessio Bechini and Eunjee Song},
pages = {1465-1474},
publisher = {ACM},
abstract = {Requirement-to-method traces reveal the code location(s) where a requirement is implemented. This is helpful to software engineers when they have to perform tasks such as software maintenance or bug fixing. Indeed, being aware of the method(s) that implement a requirement saves engineers' time, as it pinpoints the exact code region that needs to be edited to perform a bug fix or a maintenance task. Engineers produce traces manually as well as automatically. Nevertheless, traces are incomplete. This limits the amount of information that could be used by an automated technique to check further traces. Therefore, since traces are incomplete, we would like to study the effect of incompleteness on the automated assessment of requirement-to-method traces. In this paper, we apply machine learning on either incomplete or complete tracing information and we evaluate the effect of incompleteness on checking trace information. We demonstrate that the use of complete traces might yield a higher precision but yields a lower recall. Also, the use of incomplete traces yields a higher recall but a lower precision.},
bibsource = {dblp computer science bibliography, https://dblp.org},
biburl = {https://dblp.org/rec/conf/sac/HammoudiMME21.bib},
doi = {10.1145/3412841.3442021},
file = {:Conferences/SAC 2021 - On the effect on incompletness to check requirement-to-method traces/On the Effect of Incompletness to Check Requirement-to-Method Traces-preprint.pdf:PDF},
keywords = {FWF P31989, FWF P29415, Pro2Future, SCCH},
timestamp = {Mon, 03 May 2021 14:35:13 +0200},
url = {https://doi.org/10.1145/3412841.3442021},
}