Toward a Multi-Column Knowledge-Oriented Neural Network for Web Corpus Causality Mining

Round 1

Reviewer 1 Report

The authors gave a thorough state of the art.

Line 43: Would improve readability to re-phrase after: "It leaves NLP with an important and open challenge: +++"

Line 57: I would propose the authors to introduce a short discussion on causality versus correlation.

Line 142: I suggest replacing "unusual" with novel for instance, or other word.

Line 157: typo

Line 233, 256: Re-phrase so that you do not start the sentence with a reference number.

Line 274: Did not understand the sentence: "Our influence+++"

Line 303: What statistics is meant here?

Line 327: Figure 1 quality is poor, the image is blurred. It is not possible to read the labels. This is the same for Figure 2, line 473.

Line 353: Figure 3 is mentioned here, but comes much later in the document. The figure should be immediately after is referenced.

Line 366: Same for Figure 2.

Line 472: Figure 2 is based on a figure from [27] and this should be specified.

Line 390: These examples are from [27], should be explicitly said. I believe is enough to reference the paper, no need to have this text cited here.

Line 420-470: very similar to [27].

Line 471: Algorithm is from paper [27].

General comments:

- figures quality should be improved.

- difficult to see the contributions of the paper. I suggest a figure with the sketch describing existing work and parts where this paper contributes to.

 

Author Response

Special thanks to you for your good comments. We appreciate for Editors/Reviewers' warm work earnestly, and hope that the correction will meet with approval. If you have any question about this paper, please feel free to contact us. Once again, thank you very much for your comments and suggestions

Author Response File: Author Response.docx

Reviewer 2 Report

1-      There are many spelling mistakes, and the whole document has to be revised.

2-      At the end of section 2, a table listing the various ML, non-statistical, and DL approaches for mining causality with benefits and drawbacks should be added.

3-      Why is the 'wf' filter better than the convolutional filter that was learned?

4-      Many abbreviations, such as BK in line 225, are used without any explanation.

5-      Algorithm1 is too long. For better clarity, I advise the authors to divide it into three algorithms.

6-      Figure 1 is blurry.

7-      "Valuation matrices" is misspelled on page 17.

8-      Figure 4 is difficult to understand. What is the y-axis pointing at?

9-      The significance in Table 3 and Figure 5 is the same. Table 3 is enough.

10-   Authors should include a new section for results interpretation that explains the scientific justifications for better results.

11-   To better understand the behavior of the proposed model, authors must present the MCKN's performance metrics on the testing dataset.

12-   It is necessary to provide a graph illustrating the relationship between the number of epochs and model performance matrices.

Author Response

Special thanks to you for your good comments. We appreciate for Editors/Reviewers' warm work earnestly and hope that the correction will meet with approval. If you have any questions about this paper, please feel free to contact us. Once again, thank you very much for your comments and suggestions

Author Response File: Author Response.docx

Reviewer 3 Report

1. In the structure and writing of the full text, the background information in the Abstract, Introduction and Literature review is a bit too long. Please streamline the narrative, e.g. section 2.1, and if it is not relevant to the content of the article, you can pass over it to make the article more concise.

2. Please upload higher resolution images, e.g. fig1 and 4, where the text in the images is almost impossible to read. Please also pay attention to the size and stretching ratio of the images.

3. please focus on the structure and implementation details of the proposed method. for the method-related literature review, it is recommended to add it to the literature review in section2.

4. Please highlight the innovative work contributed in the full text. It is difficult to summarize the innovation and outstanding effect of the proposed method by reading the current manuscript.

The author should consider these issues, and the paper needs to improve a lot.

Author Response

Special thanks to you for your good comments. We appreciate for Editors/Reviewers' warm work earnestly and hope that the correction will meet with approval. If you have any question about this paper, please feel free to contact us. Once again, thank you very much for your comments and suggestions

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors did a good work in reviewing the paper. I encourage the authors to check the english, have noticed several typos, for instance line 130.

Author Response

We are very grateful to the reviewer for their insightful comments on our manuscript in the second round of the review process. We have been able to incorporate changes to reflect most of the suggestions provided by the reviewer. We have highlighted the changes with red color by using Track changer within the manuscript. Please see the attachment to the point response list to the reviewers.

Author Response File: Author Response.pdf