A Survey on the Use of Graph Convolutional Networks for Combating Fake News
Round 1
Reviewer 1 Report
In this paper, the authors propose a survey concerning the usage of Convolutional Neural Networks for fake news detection. The contribution is interesting and easy to follow.
- There is "no real research question" addressed in this paper. I put this in quotes because the paper is a survey paper, and it is not meant to advance knowledge in a specific research field, but, as the name says, it surveys and collects the literature on a specific research topic. This does not lower the merit of the research proposed in this paper.
2. I think the topic is original, although some surveys on more generic scenarios have already been published.
3. It covers a very specific scenario that was not covered previously. More general models were covered previously in the literature, see https://www.emerald.com/insight/content/doi/10.1108/IJICC-04-2021-0069/full/html?utm_source=rss&utm_medium=feed&utm_campaign=rss_journalLatest and https://scholarspace.manoa.hawaii.edu/handle/10125/59713.
4.I think the paper is well written and fairly clear to computer scientists.
Are the conclusions consistent with the evidence and arguments presented?
5. The methods considered are well described, but what I think is missing is a more detailed comparison between the approaches. The paper is merely a presentation of a list of approaches, but there is only a very vague comparison in the form of a table. Given the surveying nature of the paper, I would like to see a more detailed comparison, such as a comparison concerning training times and the accuracy of the models. I think that these are two crucial aspects missing in the paper because the reader reading a survey would also like to get a better understanding concerning how the surveyed approaches work in the field. For instance, some models are better suited when computation is not a limited resource, others may perform better under other circumstances. But this type of comparison, while needed (in my opinion), is not addressed with a proper level of detail in this paper. I would also recommend the authors to describe the weaknesses of each approach: the authors describe what each model can do, but not their limitations.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors have performed a review of techniques dedicated to the detection of fake news, the paper in its current form requires the following additions:
- Above all, this work should include details of the architecture of the discussed models, including conceptual drawings,
- Then the effectiveness of the different methods in the selected contexts should be discussed, with examples of results shown, results of the action,
- the codes for the individual techniques in turn, or bibliographic entries containing the codes should be linked,
- Finally, it is worth adding a section critically discussing the different techniques, what problems are encountered and to what extent, e.g. determining optimal parameters, computational complexity, difficulties in operating in real time and how to stop learning.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I think the authors did a great job towards improving the quality of the paper, and therefore I suggest accepting the manuscript.
Reviewer 2 Report
Despite the fact that the authors got the reviews mixed up and attached as a reply, a reply to another review. Seeing the current version of the paper I can see that my comments have been taken into account. I have no further comments.
