The paper in the field of privacy protection is about a proposal to improve the k-anonymous approach. Using the proposed approach, the drawback of k-anonymous that at least k elements must have the same quasi-identifier is tackled. The authors state their new k-anonymous can resist homogeneity attack, background attack, and exhaustive attack. However, these benefits are not discussed.

The paper is concise and fits its intended purpose. Honestly, I am no expert of the field of privacy protection, but I assume most readers of Applied Sciences are not either. So some of my reading experience might be true for a large portion of the intended audience.

To improve the paper, I made several comments directly in the submitted PDF document. The authors are kindly asked to consider these comments where appropriate.

So, I will focus on my main points here:

• The introduction should address some real-world contexts where the presented approach can be applied in.
• Section 2 is more a summary of K-anonymous and differential privacy. That kind of content belongs in a textbook, not in a research paper. The authors should more reflect on other/similar approaches from other researchers and should explain how their work is better/different. This submission might be a paper where the related work section is better placed before the conclusion?
• The description of the algorithm (which is decomposed into four sub-algorithms) can be optimized. I think it would be better for the reader to have an overall workflow/dataflow diagram that explains what sub-algorithm processes what kind of data.
• Some details of the algorithms seem to have spelling errors or variables like T or PHI have not been introduced systematically.
• I missed some critical discussion about the selection of algorithm details like distance functions but also about the limitations of the approach (e.g. non-numerical data without VGHT?)
• The methodological part must be improved (at least in its presentation). The authors make use of the KLD (relative entropy) distance measure to measure the information loss. In the case of privacy protection, an higher information loss seems to be preferable. However, in their discussion, a smaller information loss is rated better? That irritates me.
• What is more: In 9 use cases, the proposed approach is worse, in 6 use cases indifferent and only in three better. If we take the misinterpretation of the KLD metric into account (it would be 3, 6, 9). However, these results are quite heterogenous and should be discussed critically by the authors.
• I miss a threat of validity section. The results seem to be (at least partly) dependent on the used datasets. The authors should address strengths and weaknesses of their approach.
• I got lost in several parts of the paper. 1) Algorithm description/design 2) Comparison methodology 3) Discussion of experimental results. These parts must be improved. Detail comments can be found in the appendix.
• I missed a discussion about how this approach can be applied in real-world contexts. I think this is a vital aspect for a Journal called "Applied Sciences".
• Some grammar and typos are still in the submission. The authors may let an English native speaker do the final proof-reading or should make use of Grammar checkers like Grammarly (premium edition).

So, my recommendation would be to leave the final decision about whether to accept or reject the paper to an expert in the field of privacy protection. However, in all cases, I recommend at least a major revision for the paper.

The authors provided a medium revision of their paper considering my and other reviewers remarks (point by point). However, most of the reviewing remarks still have to be addressed. At least

• the overall readability
• the presentation of the methodology
• the presentation of algorithms
• the presentation of results
• the critical discussion if strengths/weaknesses and limitations of the approach
• the related work

This revision shows the main problem of the paper. The basic structure of the paper is not useful to transport the overall (and valuable) message of the paper. What is more, yes - the authors took their initial version and adapted it to cover reviewers remark but that did not solve the overall problem.

Let me make this more clear by taking the arguments of the other reviewer of the first round.

As it was remarked by reviewer #1, the writing style/clarity needs more effort before this work will be ready to publish. But after this revision, the paper is still difficult to read, because of the unclear structure of the argument being put across. The overall structure of the paper has not been changed. Reviewer remarks flow into the document point by point but did not change the overall structure (and therefore the overall readability) of the paper.

Furthermore, reviewer #1 remarked that the quality of the presentation significantly weakens this paper. This is another point that can not be addressed by a point-by-point revision. With this revision, it is still not clear what to expect with the direction of the article. A proper related work section is still missing. The paper's proposal is still hardly compared and contextualised with respect to the state of the art. As a result, it is still extremely difficult to understand the novelty introduced by the paper.

The authors are recommended to take a proven outline for such kind of solution proposals (Introduction, Methodology, Proposal description, Results/Critical discussion, Threats of Validity/Limitations of the approach, Related Work (comparison with different approaches, no repeating of the theory of the field), Conclusion), to take their results, take the reviewer comments and rewrite their paper completely new and from "scratch". What is more, the authors should take similar papers into account to learn how to adapt their overall structure and get impressions on how to outline the overall structure of the arguments being put across.

So, the building blocks of the paper are fine but they have to be arranged in a completely new outline. I think doing this is not possible with a point-by-point revision but needs a completely new draft of the paper. And that means a reject (by the review) and new submission (by authors). However, the authors should feel encouraged to do this.