In this paper we apply self-labeling algorithms as Semi-Supervised Classification (SSC) techniques in order to automate the classification of functional and non-functional requirements contained in reviews in the App Store. In this domain, where it is easy collect a large number of review but difficult to manually annotate then, we found that SSC techniques can successfully perform this task and that only a small amount of data is needed to achieve results similar to classical supervised techniques. We also found that the models learned can properly assign labels to the collected data and can classify unseen future reviews. We believe SSC techniques can be of particular use during requirements classification.
Deocadez, RogerHarrison, RachelRodriguez, Daniel
Faculty of Technology, Design and Environment\Department of Computing and Communication Technologies
Year of publication: 2017Date of RADAR deposit: 2017-08-11
All rights reserved
RADAR: Research Archive and Digital Asset RepositoryAbout RADAR