Title:
IR based Traceability Link Recovery Method Mining
Author(s):
Takeyuki Ueda, Shinpei Ogata, Haruhiko Kaiya, and Kenji Kaijiri
Source:
In The Eightth International Conference on Software Engineering Advances
(ICSEA 2013),
pp. 278-284, Venice, Italy, Oct. 2013. Oct. 27 - Nov. 1.
Abstract:
Traceability link recovery is an important process in
software development, and several researches are done, but the
generality of adequate methods is not considered. The target of
traceability link recovery includes several kinds of documents, so
the adequacy of recovery methods depends on the characteristics
of these documents, for example, an average similarity, a kind
of document pair, document size, and so on. We propose the
traceability link recovery method mining, which identifies a
kind of adequate recovery method based on the characteristics
of target documents by using knowledge base consisting of
(a method, characteristics, and performance). This knowledge
base shows which pair (a method, characteristics) is good at
performance. Our target traceability link recovery method is IR
based method, which is major method of automated traceability
recovery. Some experiments based on the traceability reference
data sets are done and the potential of our method is shown.
Related Paper(s):