Cost-Sensitive Multigranulation Approximation in Decision-Making Applications

Round 1

Reviewer 1 Report

The manuscript is well-organized. The authors proposed a cost-sensitive multigranulation approximation of rough sets, including optimistic approximation and pessimistic approximation. They also present a cost-sensitive selection algorithm based on the multigranulation approximation. They evaluated their approaches with six UCI datasets and reported their multigranulation approximation produces the least misclassification cost.

 

A few suggestions for this manuscript:

1) I would suggest the authors proofread the manuscript again to fix the typos. There are a notable number of typos in the manuscript, not limited to the followings:

a) In the title, "Cost-sensitive" => "Cost-Sensitive"

b) In the introduction, "GrC sloves the complex problems.." => "Grc solves the complex problems..."

c) In the introduction, "...which complex information are divided..." => "...which complex information is divided..."

d) In the introduction, the authors mention "GrC models mainly cover three types: ...", but four types are listed.

e) In the introduction, "cost sensitive" => "cost-sensitive"

f) In the introduction, "...different for each knowledge space.." => redundant periods.

g) In the introduction, "In three-way decisions model..." =>" 

 

2) The authors evaluate their approach with six complete UCI datasets, whereas the multigranulation rough set theory is especially useful for incomplete data. If incomplete data is beyond the scope of this work, I would suggest explicitly mentioning that in the manuscript; If it is in the scope, it's better to add some explanation of whether your approach will work for incomplete data.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

1. I feel that the Introduction section is far too long and therefore difficult for potential readers to follow and understand. As such, I recommend rethinking and redoing it, in addition, I think it is necessary to create a new Literature Review section in which the opinions of other authors who have published reference works for the field addressed in this paper should be presented, thus establishing the benchmarks where authors should start their research and where their original contributions to the field can be easily identified.

2. I think that in section 6. Experiments and analysis, the number of experiments and data sets used should be increased to give greater relevance and credibility to the results obtained and implicitly an increased scientific value of the present work.

3. Based on the previous recommendation, the necessary framework is created to create a new Results and Discussions section that presents, in detail, the results of the research carried out by the authors and demonstrates that the set objectives have been achieved. In addition, the necessary framework is created to expand the Conclusions section that will make the paper easier for readers to understand.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Abstract, title and references:

(1) The abstract should follow the style of structured abstracts: 1) Background: Place the question addressed in a broad context and highlight the purpose of the study; 2) Methods; 3) Results; 4) Conclusion. Abstract has come out well but it doesn’t have the statistical information such as how these experimental results have solved the real-time issue and what is the percentage error in comparison to different solutions.
(2) After going through references, it seems nearly 85-90% of the
articles are cited from a single country, it somewhere gives
unethical notion to reviewer as well as editor.
(3) Self-citation is another parameter you cannot have more than 2
articles from your earlier work (Yang, J. quoted his 4 articles), otherwise it gives wrong notions to readers as well as young researcher.

Introduction
(1) Introduction section has come out well provided there need to be a
clear substantiation on the gap or limitations in the existing
methods.
(2) Total 46 article citations are included. Out of which in
introduction it is also 46. It is commendable, many authors forget it.

Preliminaries, Cost-sensitive approximation model of rough sets,
Cost-sensitive multigranulation approximations of rough sets,
The optimal multigranulation approximation selection

(1)There is no section known as methods, do modify sections
under the head of methods or separately mention information
pertaining to methodology.
(2) Somewhere a typical case study would have been easier for understanding of the entire process.

Results
(1) Do modify section 6 under the head of Simulation Experiment and Result Analysis.
(2) Figures 4, 5 and 6 should be organized differently, it is difficult to read anything from them and the reader must make an effort to see the captions.
(3) Results and discussions are acceptable only with any correlation
between the current used material and existing materials in public
domain.

Conclusions
The discussion section lags in explanation with respect to the
work carried out.
There is no proper representation of the work with validation and
justification, hardly articles are being used in discussion section.
Conclusion looks to be generic need to compile the outcomes and
state based on the tests conducted and convey how best this can fit
in the current context.
In conclusion section, values have to be displayed with
explanation. It's better to mention the salient features of the entire
work in terms of bullet points with current context.

Overall
The topic is relevant to current context but needs to incorporate all
the modifications suggested above.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf