Analyzing Markov Boundary Discovery Algorithms in Ideal Conditions Using the d-Separation Criterion

Round 1

Reviewer 1 Report

Dear Authors

The manuscript is somehow interesting and it may possess sufficient material to be considered as a potential for publication in this journal. The language must be reviewed. The paper organization must be improved. There are major comments that must be addressed by the authors in the revised version.

1. Please avoid using the active tense starting with “we”. It is better to use the passive tense; it is more academic.
2. Abstract: please fully specify IAMB 3 and IPC-MB, since it is the first place to state.
3. What is the main novelty of the work? Its contribution to the state of the art? Please stress it in the abstract and discuss it in the introduction section.
4. Why the Authors decided to study these two algorithms amongst others?
5. The paper lacks a table of nomenclature. Please include it in the revised version.
6. Can you please discuss the computational costs?
7. How did you implement the developed algorithms? Which software?
8. It is better to present a flowchart on the step by step to show how the algorithms runs.
9. The paper lacks the methodology description and also more examples and comparison. Please comment it
10. How the accurate test values were obtained? The ones presented in tables?
11. Please rewrite the conclusions. In fact, reorganise it in the way that shorter and more comprehensive.

Very Best

The Reviewer

Author Response

Please see attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

1. Abstract should be clear and check whether all the points mentioned in abstract are addressed in this manuscript. The abstract must summarize the performance evaluation results.
2. The authors should provide solid motivation for their work based on the existing literature. In addition, the main contributions should be defined as set of bullets at the end of introduction section.
3. A table should be added to summarized related work and state the approach and result achieved by other research in the introduction.
4. The paper contains only references before 2022; other years references also need to be added especially of 2021 and 2022.
5. English usage needs to be improved. Please carefully proofread the paper to get rid of typos and grammar mistakes.
6. References should follow the template. Please correct them all.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Report on the paper titled "Analyzing Markov boundary discovery algorithms in ideal conditions using the d-separation criterion" submitted in Algorithms. 

The paper may be summarized as follows: In instances when the behavior of Markov boundary discovery algorithms must be evaluated under ideal conditions, this article recommends the use of the d-separation criteria as a conditional independence test. To demonstrate this method, implementations of two such algorithms, IAMB and IPC-MB, were set up to employ d-separation as a conditional independence test, computed on the known Bayesian networks used to synthesize the data sets. Using d-separation as a conditional independence test eliminates sources of suboptimal behavior, allowing algorithms to function at their theoretical best and provide information about their attributes. The algorithms were then set up to illustrate their behavior on synthetic data using the statistical G-test, which allowed for comparisons. This method can be applied to a variety of other situations as well. This method can also be used to create meaningful efficiency and accuracy measurements.

Evaluation: The paper has some interesting ideas, but it seriously lacks precision and substance. The major points are:

• What is the exact algorithm used, with all the details? It must be written in a professional way, using arrows and boucles, and placed under the table forma. The details are very unclear now. It cannot be validated in its present state. 
• The mathematical background seems weak. The reader does see the mathematical basis of this work. They all appear to be scholars, not researchers.
• The novelty and the importance of the findings must be clarified too. 

  For these major reasons, I cannot recommend this paper. However, after a deep revision following the points above, I can revise my evaluation.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear authors

Good improvement!

Accepted!

Best regards

The Reviewer 

Reviewer 2 Report

The authors addressed all comments 

Reviewer 3 Report

The revised version convinces me more, with its more solid background, clarification of the objective and findings, and clear algorithm. It is more professional and reaches the level of exigent research. I suggest acceptance of this revised paper.