Classification of Different Recycled Rubber-Epoxy Composite Based on Their Hardness Using Laser-Induced Breakdown Spectroscopy (LIBS) with Comparison Machine Learning Algorithms
Round 1
Reviewer 1 Report
In the Yılmaz et al.’s study, machine learning is combined with laser-induced breakdown spectroscopy to successfully detect dangerous substances in a chemical. The article content and visualization are given well enough, I suggest acceptance after minor revision.
1. The resolution of images can be improved to improve quality during printing in the journal.
2. The conclusion part can be expanded by considering strengths and weaknesses materials for different applications.
3. The graphs showing the analysis result should be interpreted with reference.
Author Response
Dear Reviewer-1
Thank you for your kind help for our correction of the manuscript
We have put here our responses as an attached file
Kind Regards
E. BAYRAKTAR
Corresponding author
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This manuscript integrated Machine Learning (ML) into Laser Induced Breakdown spectroscopy (LIBS) to distinguish five synthetic polymers and eight heavy material content. The application of ML to the characterization of materials is a hot topic. The community would find this work fascinating. However, to improve the current manuscript, I believe the following comments should be addressed:
Comments 1): The authors show that Kernel Naive Bayes Model accuracy is %99.0. How is this accuracy value determined?
Comments 2): In the experimental section, the details of the machine learning are absent. What software/language was used? How was data adjustment performed? Please elaborate. The sample preparation is in 2.1, are there 2.2 and 2.3 missing?
Comments 3): The abbreviations are not used properly. Please check the duplicated full names. For example, in the introduction, the full name of LIBS has been used multiple times after the abbreviation. Similar issues should be avoided.
Comments 4): Please reorganize Figure 4. All the sub-figures have different sizes.
Author Response
Dear Reviewer-2
Thank you for your kind help for our correction of the manuscript
We have put here our responses as an attached file
Kind Regards
E. BAYRAKTAR
Corresponding author
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper compares different ML methods to analyze the heavy material contents of plastic materials. The topic is very interesting, and a comparative study would help researchers to select the best ML algorithm for similar material analysis. However, this manuscript shows a lot of flaws, and an extensive revision is required to make this paper suitable for publication. From a general point of view, the main impression the reader has is confusion: between the title, the content of the abstract, the conclusion, and the text itself. The introduction is not well structured, the methods section is very lacunar, and more important the presentation of the design of the experiment shall be significantly improved (see comments below). The content of the paper lacks novelty (that is normal for a comparative study), I also have a concern about the choice of the Journal. To be reconsidered if the authors decided to submit again this research.
Hereafter is a detailed list of comments:
1. Introduction: to be revised, difficult to follow, notably the part that introduces the present research, please better describe the approach used here and the novelty
2. Experimental: please add section 2.1: e.g., LIBS system description, and rename 2.1 to 2.2 for Sample Preparation
3. A short introductory sentence could also be added
4. About the section “sample preparation”, it seems that some materials shall be moved to the introduction, or at least the introduction shall describe recycled rubber-epoxy composite further
5. Also, it is unclear how many samples were made, and how the raw material is obtained.
6. It is unclear why machine learning algorithms are not discussed in Section 2 (methods), please revise (move the first part of §3 in section 2)
7. Also, the algorithms shall be better described (lines 179-182). How the authors did test these algorithms? Using which software?
8. The discussion of the results (lines 138 to 153) is “confusing”, please prefer a more systematic approach to better highlight your findings.
9. Wavelength units are missing (see e.g., line 120)
10. The design of the experiment is confusing, please add schematics to better explain it
11. The title partly reflects the content of the paper, please consider adding a reference to ML.
12. Figures 4 to 14 are blurry, the legends are not sufficient, and the reader shall always refer to the text to understand the Figure (class ID…)
13. In Figure 14, the true value is difficult to find, please revise.
14. Several typos in the text, verbs are missing in some sentences (lines 33, 39, 47…), and extensive revisions are required
Author Response
Dear Reviewer-3
Thank you for your kind help for our correction of the manuscript
We have put here our responses as an attached file
Kind Regards
E. BAYRAKTAR
Corresponding author
Author Response File: Author Response.pdf
Reviewer 4 Report
This paper used the data produced by Laser Induced Breakdown Spectroscopy integrating Machine Learning to detect harmful materials. The paper has good practicality， but at this stage, the content of the paper need to be added.
1. The abstract is too short to be understood by the readers about the materials used, the ML methods used, what is the final results, what the meaning of the research.
2. The introduction part is also need to be improved about the composite research background, why we need to do the work. The chosed ML methods research background and traits, how we slected the methods. In our reserch work, why it is appropriate。
3. Figure 2. Materials of different colors need to be labeled。
4. Figure 3 may not be included in the paper in my opinion
5. Diagrams of the same structure can be composed into a large diagram for comparison and illustration.
Author Response
Dear Reviewer-4
Thank you for your kind help for our correction of the manuscript
We have put here our responses as an attached file
Kind Regards
E. BAYRAKTAR
Corresponding author
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
The paper is now better structured and could be considered for publication after editing of the English language (in particular the new descriptions added in version 2)
Reviewer 4 Report
The authors of the paper have made comprehensive revisions according to the revisions of the reviewers, and the overall quality of the paper has been greatly improved.
