Next Article in Journal
Technical-Economic Opportunities of Mixture Optimization and Recycled Aptitude at a Pre-Casting Concrete Plant
Next Article in Special Issue
Efficient Node Insertion Algorithm for Connectivity-Based Multipolling MAC Protocol in Wi-Fi Sensor Networks
Previous Article in Journal
Convergence Check Phase-Field Scheme for Modelling of Brittle and Ductile Fractures
 
 
Article
Peer-Review Record

Leveraging Graph-Based Representations to Enhance Machine Learning Performance in IIoT Network Security and Attack Detection

Appl. Sci. 2023, 13(13), 7774; https://doi.org/10.3390/app13137774
by Bader Alwasel 1,*, Abdulaziz Aldribi 2, Mohammed Alreshoodi 1, Ibrahim S. Alsukayti 2 and Mohammed Alsuhaibani 2
Reviewer 1:
Reviewer 2:
Reviewer 3: Anonymous
Appl. Sci. 2023, 13(13), 7774; https://doi.org/10.3390/app13137774
Submission received: 15 May 2023 / Revised: 25 June 2023 / Accepted: 28 June 2023 / Published: 30 June 2023
(This article belongs to the Special Issue Advances in Wireless Sensor Networks and the Internet of Things)

Round 1

Reviewer 1 Report

Overall, the paper has strength and presents current research trends. The authors highlighted issues regarding machine learning models based on graph theory to elevate their anomaly detection capabilities in the context of Industrial Internet of Things (IIoT) network data analysis. There are the following issues required to handle by authors.

The abstract is clear; firstly, it presents a context about the environment in which the study is useful, then it presents the problems that arise when there is no complete knowledge about the application environment, and finally, it places the study as a clear and efficient alternative for know the various concepts involved; however, it would be useful for the reader to mention in the abstract that a more in-depth description.

Section 3, Proposed Approach, Added a flowchart with a step-by-step procedure, which helps to reader better understand the proposed approach. Also, the authors said that (lines 191 to 194) in their approach, they experiment with logistic regression, support vector machines (SVM), and k-means clustering, assessing their performance in terms of accuracy, precision, recall, and F1 scores, by comparing the classification performance of these models before and after applying graph theory representation. But I did not find any information about logistic regression, support vector machines (SVM), and the k-means clustering approach section; what is the main significant mention of classification? 

Section 6.2, Feature Analysis Before Graph Theory Representation. Added more explanation on figure 8 & 9 result.

The conclusion should be exact and linked with the abstract. 

Other than that, the paper quality is good, and the research is interesting. 

Moderate editing of the english language required

Author Response

We would like to convey our sincere gratitude to reviewer 1 and the editor for their time taken to review our paper and the constructive comments. We have carefully modified the paper to reflect all the comments/questions raised by the reviewers. Below we summarise how each comment was addressed and provide pointers to the changes made in the revised version of the paper. To improve readability, we use blue colour to show our responses, whereas the comments from the reviewers are shown in black.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this paper, the authors investigate the underexplored potential of fusing graph theory with machine learning models to elevate their anomaly detection capabilities in the context of Industrial ΙοΤ network data analysis. A list of points that appears to deserve to be better clarified in the paper together with some suggestions follows.

§  More details are needed to clarify the structure of the machine learning models in section 6.3.

§  The way that the graph theory is applied to network data should be described in more detail.

§  In section 6.4, the comment regarding “These figures clearly demonstrate the improvement in classification accuracy, highlighting the significance of graph theory representation in the context of network data analysis.” should be better explained.

§  The authors should refer in more detail and in clearly way to the advantages of their methodology and summarize possible limitations.

§  The reader of this article will be interested in the articles:

-           Manjunath A., Neelappa, Prakash, Veeramma Yatnalli, Saroja S. Bhusare, "Performance Analysis of Graph theory-based Contrast Limited Adaptive Histogram Equalization for Image Enhancement," WSEAS Transactions on Systems, vol. 22, pp. 219-230, 2023.

-           Ridho Alfarisi, Liliek Susilowati, Dafik, Osaye J. Fadekemi, "On The Local Multiset Dimension of some Families of Graphs," WSEAS Transactions on Mathematics, vol. 22, pp. 64-69, 2023.

-           Milos Seda, "Steiner Tree Problem in Graphs and Mixed Integer Linear Programming-Based Approach in GAMS," WSEAS Transactions on Computers, vol. 21, pp. 257-262, 2022.

Author Response

We would like to convey our sincere gratitude to reviewer 2 and the editor for their time taken to review our paper and the constructive comments. We have carefully modified the paper to reflect all the comments/questions raised by the reviewers. Below we summarise how each comment was addressed and provide pointers to the changes made in the revised version of the paper. To improve readability, we use blue colour to show our responses, whereas the comments from the reviewers are shown in black.

Author Response File: Author Response.pdf

Reviewer 3 Report

Section 2:  repeated formulations, needs a better structure (maybe a table / subheadings). Also, you do not use very cited papers like "Graph-Based Anomaly Detection" by Noble & Cook, or have any discussion on the pros and cons of Graph Based Models. No mention of GNN, which work beautifully with complex networks.

Section 3: describe how do you handle  missing values, outliers, and noise? how do you standardize? how much actual data loss using PCA? why did you choose those 3 ML methods? why are they relevant?

Section 4: why use only a star topology? why not mesh/ring/bus /etc?

Section 5: one dataset you give numerical characteristics, the other not. Needs a uniform approach

Section 6: PCA: too little points to draw that conclusion. Figures 10 and 11 can be condensed, numerical value of accuracy should be used to showcase accuracy percentage. What does it mean compared to the dataset size? Analysis time increase by using CN? Comparison with other methods (even if in different fields)? Mathematical support? Can benefit from more experimental data.

Conclusions: a bit oversold.

Try to use more academic formulations. For example "As we forge ahead in the realm of network security, it is imperative to continuously explore and innovate in order to fortify our defenses against the ever-evolving landscape of cyber threats." might benefit from some more impersonal and less dramatic change in style.

Author Response

We would like to convey our sincere gratitude to reviewer 3 and the editor for their time taken to review our paper and the constructive comments. We have carefully modified the paper to reflect all the comments/questions raised by the reviewers. Below we summarise how each comment was addressed and provide pointers to the changes made in the revised version of the paper. To improve readability, we use blue colour to show our responses, whereas the comments from the reviewers are shown in black.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The respected author overcame all my recommendations given in the previous review report.

No more further recommendations.

Some sentences in this article are tough to understand. Moderate editing of the English language is required.

Author Response

We would like to convey our sincere gratitude to reviewer 1 and the editor for their time taken to review our paper and the constructive comments. We have revised the English language as much as possible.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript can be published in its present form.

Author Response

We would like to convey our sincere gratitude to reviewer 2 and the editor for their time taken to review our paper and the constructive comments.

Author Response File: Author Response.pdf

Reviewer 3 Report

Section 6 much improved.

I still think you need to do a comparison with at least some of the papers mentioned in section 2, otherwise your method, no matter how novel, cannot be integrated into a wider picture.

Experiment wise it has a too narrow focus. Variation in dataset size, analysis time in the complex network approach, maybe a GNN approach VS your approach are some ideas to consider adding.

Author Response

We would like to convey our sincere gratitude to reviewer 3 and the editor for their time taken to review our paper and the constructive comments. We have carefully modified the paper to reflect all the comments/questions raised by the reviewers. Below we summarise how each comment was addressed and provide pointers to the changes made in the revised version of the paper. To improve readability, we use blue colour to show our responses, whereas the comments from the reviewers are shown in black.

Author Response File: Author Response.pdf

Back to TopTop