Genova, Italy Italy
Mar. 5, 2013 to Mar. 8, 2013
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/CSMR.2013.35
Various studies indicate that having concise and consistent identifiers improves the quality of the source code and hence impacts positively source code understanding and maintenance. In order to write concise and consistent identifiers, however, developers need to have some knowledge about the concepts captured in the source code and how they are named. Acquiring such knowledge from the source code might be feasible only for small systems, while it is not viable for large systems. In this paper, we propose an automated approach which exploits concepts and relations automatically extracted from the source code to suggest identifiers. The suggestion is ranked based on the context in which a new identifier is introduced and it can be used either to complete the identifier being written or to replace it with a more appropriate one. To validate the proposed approach, we have conducted a case study by simulating the activities of a developer in naming identifiers. The results of the study show that in the majority of the cases our approach provides completion suggestions which match the identifiers actually used by the developers.
ontology, identifier quality, source code maintenance
Surafel Lemma Abebe, Paolo Tonella, "Automated Identifier Completion and Replacement", CSMR, 2013, 2011 15th European Conference on Software Maintenance and Reengineering, 2011 15th European Conference on Software Maintenance and Reengineering 2013, pp. 263-272, doi:10.1109/CSMR.2013.35