Detecting Sudden Drift for Single-Instance Log in Software Development Process

The current algorithm for detecting drift in the business process can only detect the drift of the log of multi-instance. However, the log generated in the software development process is mainly single-instance. Hence, this paper proposes an algorithm for detecting sudden drift of the log of single-instance. First, we apply the LDA topic model for the topic extraction of log in the software development process to obtain the log of single-instance. Then the location of the drift point of the single-instance log can be detected according to the change of the lowest frequency of direct successor relation. Finally, the location numbers of the actual drift points were obtained by corresponding to original data. Multiple sets of simulated data evaluate this algorithm. The results show that this algorithm can effectively detect the sudden drift location of the single-instance log in the software development process.