The Fast Detection of Abnormal ETC Data Based on an Improved DTW Algorithm

Round 1

Reviewer 1 Report

Summary: In this paper, the authors proposed a SegrDTW algorithm, which transforms the ETC transaction data into a series of Origin-Destination (OD) paths and quickly detect the missed transactions and false transactions. Compared with the existing algorithms, the time performance of SegrDTW is improved by an order of magnitude while maintaining high accuracy.

 

Comments:

1 . The main contribution of this work seems to be the improved SegrDTW algorithm which introduces a dynamic search matching step length for the ETC data. Yet the comparison of SegrDTW and the original DTW algorithm is not clearly presented. I would suggest the authors put two algorithms together or show the difference more directly when introducing the SegrDTW algorithm.

 

2. For the complexity analysis of the algorithm, the authors should also present the complexity of the standard DTW algorithm as a baseline.

 

3. For the experiments part, the authors compared the running time of SegrDTW with other baselines yet did not compare the abnormal detection performances with others. I would suggest the authors also compare them in Table 5 for both missing transactions and False transactions. 

 

Minor comments:

 

1 . Typo Line 279: should be section 3?

 

Author Response

Dear Reviewer:

Thank you for your precious comments and advice. Those comments are all valuable and very helpful for revising and improving our paper. We have studied comments carefully and have made correction which we hope meet with approval. The main corrections in the manuscript and the responds to the reviewer’s comments are as following:

Response to the reviewer's comments:

Q1. The main contribution of this work seems to be the improved SegrDTW algorithm which introduces a dynamic search matching step length for the ETC data. Yet the comparison of SegrDTW and the original DTW algorithm is not clearly presented. I would suggest the authors put two algorithms together or show the difference more directly when introducing the SegrDTW algorithm.

Response: We add the characteristics and shortcomings of DTW in lines 337 to 367 of the manuscript. Based on the shortcomings of DTW, we add a description to the improvement of DTW.

Q2. For the complexity analysis of the algorithm, the authors should also present the complexity of the standard DTW algorithm as a baseline.

Response:  In line 500 to 507 of the manuscript, we add the time complexity analysis of standard DTW based on ETC data.

Q3. For the experiments part, the authors compared the running time of SegrDTW with other baselines yet did not compare the abnormal detection performances with others. I would suggest the authors also compare them in Table 5 for both missing transactions and False transactions. 

Response: We have revise Table 5 in line 687 to 715 of the manuscript, which supplements the comparison between SegrDTW and DTW in precision, accuracy, recall and F1-score.

Q4. Typo Line 279: should be section 3?

Response: Yes, we have changed section 2 to section 3

Reviewer 2 Report

Overall paper seems good to read and scientifically better. The author mentioned mathematical works with sufficient explanation.

1.       Title of the paper should be improved and more precise because it seems not fit as per the paper scope.

2.       I suggest that, if possible, then the author should add a figure in introduction related the topic.

3.       The abstract must be rewritten as per paper's working

4.       Over all paper's grammar must improve. Such as " Aiming at the characteristics of ETC transaction data, such as OD data constraint, coupling between gantry path and vehicle trajectory, etc., this algorithm uses end-to-end bidirectional matching to extract dynamic step length and construct DTW search space window, which reduces the time consumption of matrix calculation in traditional DTW algorithm”.

 

5.       Also need to update the conclusion.

Author Response

Dear Reviewer:

Thank you for your precious comments and advice. Those comments are all valuable and very helpful for revising and improving our paper. We have studied comments carefully and have made correction which we hope meet with approval. The main corrections in the manuscript and the responds to the reviewer’s comments are as following:

Response to the reviewer's comments:

Q1. Title of the paper should be improved and more precise because it seems not fit as per the paper scope.

Response: We have changed the title to “A Fast Detection Algorithm of ETC Abnormal Data Based on Improved DTW Algorithm”.

Q2. I suggest that, if possible, then the author should add a figure in introduction related the topic.

Response: This manuscript provides the frame diagram in line 315 of the manuscript.

Q3. The abstract must be rewritten as per paper's working

Response: We rewrite the abstract according to per paper's working in line 12 to 25 of the manuscript.

Q4. Over all paper's grammar must improve. Such as " Aiming at the characteristics of ETC transaction data, such as OD data constraint, coupling between gantry path and vehicle trajectory, etc., this algorithm uses end-to-end bidirectional matching to extract dynamic step length and construct DTW search space window, which reduces the time consumption of matrix calculation in traditional DTW algorithm”.

Response: We have invited a native English speaker to proofread this paper and made the corresponding revisions to improve the overall readability.

Q5. Also need to update the conclusion.

Response: We also rewrite the conclusions in line 717 to 726 of the manuscript.

Author Response File: Author Response.docx