Analyzing the Spread of Misinformation on Social Networks: A Process and Software Architecture for Detection and Analysis
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
1The authors have designed a process of social network analysis consisting of seven modules to analyze the spread of misinformation during the COVID-19 pandemic. The experimental study and results have been reported on the implementation. However, there are some issues as follows.
11. Line 28: However, most models that examine the propagation of information and knowledge in social networks do not distinguish between different types of information, including misinformation, and disinformation.
– Can you add references for this sentence?
22. Line 67: Studies on misinformation have primarily focused on detecting [8], [9] and/or preventing false information on the internet and social media, while studies that consider the spreading factors of misinformation [10] are less.
—are both [8] and [9] detecting false information? Any references for preventing? Please Put a reference in the preventing part.
33. Line 146-150: [25],[26][27] can you put each reference for each dataset (in each sentence) separately?
44. Line 317: 2022 or 2021?
55. Readers might have interests on the final report of Misinformation Detection Module. Fig 7 just shows “report and module are finished”. Can you have a screenshot of the output report?
66. Section 6 is “Prototype & Evaluation”. I really don’t know how to evaluate after reading section 6. Authors also mention evaluation in the future work. However, I think evaluation is necessary and makes the paper more convincing.
Comments on the Quality of English Language
Minor issues:
1. Line 36: a space before (
2. Line 41: We aim to use the developed prototype application to detect how misinformation spreads on the Twitter social media platform and generalize it to be used over a variety of SNA datasets. .
Line 45: Furthermore, the findings obtained within the scope of this research provide significant insights into how misinformation spread can be analyzed on Twitter, aiming to test it on other datasets from other social networks in the future.
Can you please Rephrase these two sentences to make them more readable?
3. Line 109: In connection with the aforementioned however, we want to give an important message, that when analyzing social network data, the minimum of data points should be utilized, only the necessary to answer research questions, and this should be done with Privacy by Design, and GDPR (or other privacy-related laws outside of the EU) in mind as the minimum guidelines for privacy considerations.
-- “The aforedmentioned literature, however,.
-- give an important message, that – remove , the following sentence is too long. Can you Rephrase?. Maybe you can separate into sentences.
4. Line 147: Missing a space before ReCOVery
Author Response
We thank the reviewer for hel** us improve the paper's quality. We addressed the reviewer's comments and updated the manuscript according to the comments and suggestions made by the reviewer. We added an Author's Notes File and explained what we did for each suggestion.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for Authors
1. The authors have attempted to identify the spread of misinformation through the Twitter fake accounts. In fact, the focus of the authors is more on fake account identification and Twitter social media application, both of which have been explicitly highlighted by the authors too. In wake of these facts, the title of the manuscript seems to be a bit too general. It is suggested that the title be rephrased to be more closer and accurate representative of what is being presented rather than design in a generic way.
2. The corresponding long forms should accompany all the ‘first usages’ of abbreviations (in the abstract and the remaining manuscript). At the same location, the words of the long form should be suitably written in Title Case. Either the style of ‘long form followed by the abbreviation’ (preferably) or the ‘abbreviation followed by the long form’ should be consistently used throughout the manuscript. After the abbreviation has been defined in the first instance, the subsequent text of the manuscript should not unnecessarily mention the abbreviation and long-form again, and rather only the abbreviation should be used.
3. All the figures with graphical content need to have captions for ‘both’ axes in addition to the labels for the axes.
4. Overall, both the scope of the research work as well as the novelty of the research work is limited. The authors can explicitly present a list of specific novel contributions to the scientific community.
5. The proposed work is not the only of its kind. It is highly recommended that the authors present a ‘table’ of comparison of the proposed work with similar research works and state-of-the-art research studied as part of the literature review. This will go a long way in emphasizing the research gaps and highlighting the specific contributions of the proposed work. Irrespective of the presented discussion, this should be done in terms of contrast and comparison of approach as well as the contrast and comparison of the results.
6. What is difference between Reference No. 34 and 35?
7. What is difference between Reference No. 36 and 37?
8. Fig. 7 is not found to be cited in the manuscript.
9. Why Fig. 5 is numbered and placed in the manuscript before Fig. 4?
10. It is recommended that the authors search for ‘misinformation saini barve’ to find many related research papers dealing with misinformation detection in the healthcare domain.
11. It is strongly recommended that a legend be included with Fig. 6.
12. It is strongly recommended that the reference to folders and files be restricted in the manuscript.
Comments on the Quality of English Language
Acceptable.
Author Response
We thank the reviewer for hel** us improve the paper's quality. We addressed the reviewer's comments and updated the manuscript according to the comments and suggestions made by the reviewer. We added an Author's Notes File and explained what we did for each suggestion.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
The paper's objective is to identify misinformation and disinformation within the Twitter environment, proposing the development of software capable of transforming datasets into user-friendly applications that focus on central nodes, communities, and the detection of misinformation and disinformation. However, the paper falls short in several aspects of its academic rigor, presenting numerous flaws that hinder its recommendation for acceptance.
(1) In the “related work” section, the paper lacks an exhaustive review of existing research on misinformation and disinformation detection. Prominent scholars in this field, such as Prof. Filippo Menczer, have extensively addressed this issue over the past five years. The inclusion of these researchers' contributions is vital for situating the paper within the broader context of social network analysis.)See, for example: Pierri, F., DeVerna, M. R., Yang, K. C., Axelrod, D., Bryden, J., & Menczer, F. (2023). One Year of COVID-19 Vaccine Misinformation on Twitter: Longitudinal Study. Journal of Medical Internet Research, 25, e42227.)
(2) Additionally, the related work section introduces methodological issues related to privacy and data privacy, which are not subsequently addressed in the paper. This omission raises questions about the section's relevance. Furthermore, the paper briefly touches upon significant social network analysis terminology, like degree centrality and closeness centrality, which requires more comprehensive coverage.
(3) A critical shortcoming is the failure to acknowledge the directed graph nature of the Twitter platform, where the direction of links between nodes holds substantial importance. In-degree and out-degree centrality distinctions (and In-Closeness and out-closeness, which are less familiar but very important) were neglected, which is essential when proposing a tool for Twitter platform analysis.
(4) The authors' choice of the "Girven-Newman" community detection algorithm is unexplained, leaving readers in the dark about its selection over other available algorithms.
(5) The paper lacks a robust explanation of the link between the identification of false tweets (misinformation or disinformation) and the classification of users who create these tweets, especially malicious users. This complex task of distinguishing users, potential spammers, and spammers on Twitter and the relation to fake/ misinformed/disinformed tweets is inadequately addressed.
(6) The Figures in the paper, specifically network maps, lack sufficient explanations, leaving readers without a clear understanding of their significance The authors seem to have created a pipeline of Python libraries without producing a real (academic or practical) added value. Existing applications offer superior insights and predictions.
(7) The paper fails to connect mathematical network parameters to real-world implications.
The absence of innovative findings renders it unsuitable for publication. It lacks a profound comprehension of Twitter's social network analysis and lacks original insights. To merit publication, a deeper understanding of the Twitter environment, social network analysis, and innovative conclusions based on a comprehensive literature review are imperative. Consequently, I cannot recommend accepting this paper for publication in “Computers” journal.
Author Response
We thank the reviewer for hel** us improve the paper's quality. We addressed the reviewer's comments and updated the manuscript according to the comments and suggestions made by the reviewer. We added an Author's Notes File and explained what we did for each suggestion.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for Authors
This paper presents a process design that utilizes social network data, including retweets, mentions, and replies, to study the spread factors of misinformation and the involvement of non-trusted or fake accounts. While the paper shows promise, there are areas that can be improved before acceptance. Here are some suggestions for careful manuscript revision:
1. Abstract Structure: The abstract provides a good overview of the paper's objectives and contributions. However, it could benefit from a clearer structure, with distinct sections for the problem statement, methodology, results, and conclusion. This will enhance its readability and comprehension.
2. Introduction Clarity: In the introduction, clearly state the existing challenges and the authors' previous work, and make use of the research contribution section. A well-defined problem statement and articulation of the paper's contributions are essential for reader engagement.
3. Methodology Differentiation: Explain the key differences between your proposed method and attributed graph embedding. Highlight what distinguishes your approach from existing methods, providing a clear understanding of the novel contributions.
4. Parameter Analysis: It's important to analyze how the parameters of the proposed methods are set within the framework. Discuss whether there is an optimal choice for these parameters. Providing insights into parameter selection will enhance the paper's practicality.
5. Figure and Table Captions: Expand the captions for figures and some tables to make them self-explanatory. This will help readers understand the content without needing to refer to the main text.
6. Experimental Details: In the Experiment section, provide more details about how experiments were conducted, including the tools and software used in the experimental setup. Transparency in the experimental methodology is crucial for reproducibility.
7. Property Preservation: Explain which properties were preserved and which properties were missed in the proposed method. This will provide a clearer understanding of the method's capabilities and limitations.
8. Comparative Analysis: It would be beneficial to compare your proposed method with more recent approaches in the field. This comparative analysis will provide a better perspective on the strengths and weaknesses of your method and help position it in the current research landscape.
9.The Literature citation is not adequate, and the related work to complex network should be discussed:
1. DAC-HPP: deep attributed clustering with high-order proximity preserve
2. Community Detection Algorithms in Healthcare Applications: A Systematic Review
Author Response
We thank the reviewer for hel** us improve the paper's quality. We addressed the reviewer's comments and updated the manuscript according to the comments and suggestions made by the reviewer. We added an Author's Notes File and explained what we did for each suggestion.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for Authors
looks good after the revision.
Reviewer 2 Report
Comments and Suggestions for Authors
The suggestions of the previous review round have been well considered by the authors.
