Correctness of data mined from CVS
Proceedings of the 2008 international working conference on Mining software repositories
An approach to software evolution based on semantic change
FASE'07 Proceedings of the 10th international conference on Fundamental approaches to software engineering
Is it dangerous to use version control histories to study source code evolution?
ECOOP'12 Proceedings of the 26th European conference on Object-Oriented Programming
| Hi-index | 0.01 |
Recently, there is an increasing research interest in mining versioning information, i.e. the analysis of the transactions made on version systems to understand how and when a software system evolves. One particular area of interest is the identification of move operations as these are key indicators for refactorings. Unfortunately, there exists no evaluation which identifies the quality (expressed in precision and recall) of the most commonly used detection technique and its underlying principle of name identity. To overcome this problem, the paper compares the precision and recall values of the name-based technique with two alternative techniques, one based on line matching and one based on identifier matching, by means of two case studies. From the results of these studies we conclude that the name-based technique is very precise, yet misses a significant number of move operations (low recall value). To improve the trade-off it is worthwhile to consider the line-based technique since it detects more matches with a slightly worse precision, or to use the number of overlapping identifiers when combined with an additional filter.