Translating Workflow Nets to Process Trees: An Algorithmic Approach
Round 1
Reviewer 1 Report
Thank you for giving me the possibility of reviewing this paper. I hope the authors find my comments productive and that it will help them to improve their research work.
In this article, the authors present an algorithm that detects whether a WF-net corresponds to a process tree, and, if so, constructs it. The authors prove that, when the algorithm finds a process tree, the language of the process tree is equal to the language of the original WF-net. The experiments conducted show that the algorithm’s corresponding implementation has a quadratic time complexity in the size of the WF-net. Furthermore, the experiments show strong evidence of process tree re-discoverability.
The subject of the study is interesting. However it would be necessary to explain why such algorithm is necessary in the field and support this explanation with further references to previous work.
The method and scientific soundness are appropriate
The authors need to improve the conclusions making reference to possible implications of your work both related to researchers and practitioners.
Author Response
Dear Reviewer,
We would like to thank the Reviewer for providing us with your valuable feedback on our article.
We appreciate the fact that the Reviewer judges our work and scientific soundness as being appropriate.
We summarize the main issue raised by the Reviewer related to the explicit motivation of our work, i.e.,:
The subject of the study is interesting. However it would be necessary to explain why such algorithm is necessary in the field and support this explanation with further references to previous work.
We agree with the Reviewer that, particularly at the beginning of the article, our work's motivation could be strengthened.
Hence, in the introduction, we have added a reference to a recently accepted article that directly computes alignment approximations on process trees.
We have also added the use cases i) process model comparison (new) and ii) process model translation (already covered in the related work section).
The changes are highlighted in red in the new version of the manuscript.
Similarly, w.r.t. the comment: The authors need to improve the conclusions making reference to possible implications of your work both related to researchers and practitioners.
We have added an explicit sentence describing the potential applications in the conclusions, i.e.:
"The contribution enables a wide variety of applications, e.g., improved computation performance of commonly used process mining artifacts, process model comparison and process model transformation."
We have refrained from referring to specific works here as we have already done so in the newly added paragraph in the introduction.
We hope that these comments and changes satisfy the issues raised by the Reviewer.
Sincerely,
The Authors.
Reviewer 2 Report
The paper is generally fine but some minor problems should be modified as follows:
- Figure 1 (a) should be represented as another way to be clearly viewed by readers otherwise it's useless.
- In Figure 13, the time performance suddenly drop on the Size 200. I don't think it's a rational results and may result in the wrong equation of the line trend. I think the authors should explain the situation.
In general, I think the paper could be accepted.
Author Response
Dear Reviewer,
We would like to thank the Reviewer for providing us with your valuable feedback on our article.
We appreciate the fact that the Reviewer judges our work to be suitable for acceptance.
W.r.t. the Reviewer’s main concerns:
- Figure 1 (a) should be represented as another way to be clearly viewed by readers otherwise it's useless.
Indeed, the activity labels in the figure are not readable in the model depicted. However, we have chosen to show the models, as i) the models are based on real data, and, ii) the example does show, i.e., despite the readability issue, that process trees contain less elements. We feel that using a smaller model, e.g., based on dummy data would not be such a strong example. Similarly, using 2 pages, would take too much space. However, to accommodate the Reviewer’s concern, we have explicitly added a footnote discussing that the main intent of showing the figure is to show the structural difference between the two modeling formalisms. - In Figure 13, the time performance suddenly drop on the Size 200. I don't think it's a rational results and may result in the wrong equation of the line trend. I think the authors should explain the situation.
Indeed, the experimental results show more fluctuations around the final data points on the x-axis of the chart. Due to the way in which the models were generated, fairly little models were obtained of that size. Hence, the number of measurements at these data points is lower. This causes fluctuations at the RHS of the chart. Clearly, if more experiments are run with models of the same size, the chart would smoothen out.
We have added a footnote in the paper describing this phenomenon. Similarly, w.r.t. the comment: The authors need to improve the conclusions making reference to possible implications of your work both related to researchers and practitioners.
We hope that these comments and changes satisfy the issues raised by the Reviewer.
Sincerely,
The Authors.
