Profiling Public Transit Passenger Mobility Using Adversarial Learning
Round 1
Reviewer 1 Report
Although the authors present a methodological framework for analyzing the mobility of public transit passenger mobility using adversarial learning, the results presented in the manuscript are not convincing or informative. In addition, the findings and conclusions of the study do not inform recommendations for public transit planning or policy development in the study area.
(1) Figure 7 shows the results of the visual analysis on a weekday, but the visualization of the result is very poor and it is not possible to obtain the results elaborated in the manuscript based on Figure 7.
(2) The manuscript lacks the necessary detailed description of all the research data, such as the temporal coverage and spatial coverage of the data.
(3) All equations in the manuscript do not provide the necessary explanations of the variables.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This article addresses the problem of public transit mobility. The authors use a frequent itemset extended apriori algorithm for detecting frequent transit patterns and also a GAN-based architecture to profiling mobility. The experiments were performed on a real dataset from Schenzhen, China. The article is interesting and well written.
May I address some issues in order to improve the article:
1) Why not choosing a transformer-based architecture (ex. GPT-2, Falcon, T5) to run the experiments? It would be important to compare the results of the proposed GAN architecture with a transformer one.
2) All figures and algorithms must be detailed on text.
There are minor typos such as:
line 201: "which generate" should be "which generates"
lines 278 and 279: there is a missing space between "namely" and "mm"
line: 289: "spatial-temporal" should be "spatiotemporal"
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The methodology proposed in the paper is quite complex. It is not clear to me which parts of the method are new and original and which are taken from the literature. I suggest the Authors explicitly pinpoint their proposed novelties.
I also get an impression the presentation of the proposed methodology is quite abstract. In my opinion the use of some examples would help in understanding the proposed approach.
Also, a brief discussion of the proposed approach in the last part of the paper is not fully convincing as to the quality of the approach.
Some numerical parameters assumed in the paper are not explained (e.g. in lines 194, 368).
Some minor language errors can be found in the manuscript, e.g. in lines 95, 197, 221, 234, 273.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have addressed my comments.
Reviewer 2 Report
I have already aproved this article in the last review round.
