Next Article in Journal
Consistency of Central and Regional Planning in the Agricultural Sectors and the Factors Affecting It in Indonesia
Previous Article in Journal
Evaluation of the Perception and Experience of Rural Natural Landscape among Youth Groups: An Empirical Analysis from Three Villages around Hefei
 
 
Article
Peer-Review Record

A Two-Stage Deep-Learning Model for Link Prediction Based on Network Structure and Node Attributes

Sustainability 2022, 14(23), 16299; https://doi.org/10.3390/su142316299
by Peng Liu, Liang Gui *, Huirong Wang and Muhammad Riaz
Reviewer 1:
Reviewer 2:
Sustainability 2022, 14(23), 16299; https://doi.org/10.3390/su142316299
Submission received: 13 November 2022 / Revised: 4 December 2022 / Accepted: 5 December 2022 / Published: 6 December 2022

Round 1

Reviewer 1 Report

I am glad to review this paper on two-stage deep-learning model, it deserves to be published. Here are some  comments:

1.The advantages and disadvantages of two-stage deep-learning model for link prediction (TDLP) should be added in Related work.

2.In 3.1, (1) and (2) are missing, and the language should be simple and easy to understand, such as "However, still there is scarce literature ..." in line 57 can be revised in the language order.

3. After listing the basic information of the four real-world networks, you should introduce the differences to express your reasons for choosing them.

Author Response

Dear anonymous reviewer,

Thank you for your comments concerning our manuscript entitled “A two-stage deep-learning model for link prediction based on network structure and node attributes”. These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have carefully studied the comments and made corrections in the hope of approval. The changes in the manuscript are highlighted in blue. In the file labeled “Response to Reviewers”, we provide a point-to-point response to the comments with references to the changes made in the text for convenience to check.

Author Response File: Author Response.pdf

Reviewer 2 Report

The scientific quality of the paper has to be improved with a better exploration of the contributions of this paper compared to previously published work in the field. A discussion of the results is lacking.

Improvement suggestions:

- Provide more than reference to support this observation “Although some scholars have begun to explore link prediction methods that integrate network structure and node attributes (e.g., Ref. [15]), the work in this area is still relatively insufficient.”

- The literature review about the topics of traditional methods and deep learning methods need to be significantly expanded. Relevant literature in the area is ignored.

- Authors note “In short, the traditional method based on similarity is widely used due to its simplicity, and it is also more effective in some real networks.” It is not clear the characteristics of these real networks.

- Early Fusion model needs to be characterized.

- A better comparative analysis of the main methods that can integrate nodes’ attributes information in deep learning should be provided.

- Regarding the experiments, authors note “The proposed TDLP method is evaluated on the real network from four different fields, …” Which are the criteria to choose these four fields?

- Section 4.4. Experimental results should be replaced by the section 5. Experimental results and discussion.

- In fact, authors discuss only technically the results. In a scientific paper it is crucial to discuss them and compare with the literature in the field. This last task was not performed.

- Authors should clarify and explore better the theoretical and practical contributions of their work.

- The number of references is low. It must be improved.

Author Response

Dear anonymous reviewer,

Thank you for your comments concerning our manuscript entitled “A two-stage deep-learning model for link prediction based on network structure and node attributes”. These comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have carefully studied the comments and made corrections in the hope of approval. The changes in the manuscript are highlighted in blue. In the file labeled “Response to Reviewer”, we provide a point-to-point response to the comments with references to the changes made in the text for convenience to check.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I recommend minor improvements mainly to improve the discussion section. Authors note: “Experiments on real datasets show the effectiveness of the model. Hence, the TDLP method not only supplements and enriches the existing work, but also provides a new research perspective for link prediction based on deep learning.”

This view is incomplete and shows that the authors focus excessively on discussing their model experiment without actually performing a comparative discussion analysis. A comparative analysis is missing here since other models may also show relevant levels of effectiveness.

Author Response

Dear anonymous reviewer,

Thank you for your comments concerning our manuscript entitled “A two-stage deep-learning model for link prediction based on network structure and node attributes”. These comments are all valuable and very helpful for further improving our paper. We have carefully studied the comments and made corrections in the hope of approval. The changes in the manuscript are highlighted in blue. In the file labeled “Response to Reviewers”, we provide a point-to-point response to the comments with references to the changes made in the text for convenience to check.

Author Response File: Author Response.pdf

Back to TopTop