Affiliations: Department of Computer Science, Ryerson University, Toronto, ON, Canada
Correspondence:
[*]
Corresponding author: Mubarak Alrashoud, Department of Computer Science, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, Canada. E-mail:mubarak.alrashoud@ryerson.ca
Abstract: The release planning process concerns with assigning
requirements to the different future releases of the software. This paper
considers three factors that govern the release planning process: stakeholders'
satisfaction, risk, and availability of resources. All of these factors
depend on human knowledge, which is always incomplete, imprecise, and
approximated. This classifies release planning as an under-uncertainty
decision-making problem. This paper proposes a prioritization approach for
generating a release plan for the next release of the software. The proposed
approach employs a fuzzy inference system engine in order to tackle the
uncertainty in the release planning process. The artifacts of the fuzzy
inference (FIS) process (the membership functions and the IF-rules) are
constructed using adaptive network-based fuzzy inference system (ANFIS).
ANFIS helps to reinforce the human knowledge with the knowledge obtained
from the historical data. Experiments show that the outputs of the proposed
framework are affected by the reliability, accuracy, and the orientation of
the historical data used to train the ANFIS module. For example, when
training the ANFIS module using data that concentrates on the factor of
stakeholders' satisfaction, the proposed framework has shown very good
results from the perspective of this factor.
Keywords: Software release planning, applications of fuzzy inference system, applications of ANFIS, fuzzy decision making
