Social Media and the Scourge of Visual Privacy
Round 1
Reviewer 1 Report
In this article, the authors analysed the users’ perspective of privacy based on the survey of 250 people and data collected from Twitter social network.
Comments:
Lines 49-57: formulate your aims as research questions (hypotheses), which the results of the survey should confirm or reject.
Line 52: check question marks.
Section 3.1: were the survey questions validated and how?
Section 3.2: provide more details on how your have preprocessed the survey responses. Did you remove common words? Did you apply Porter stemming etc.?
Provide a dendrogram to show the relationship between clusters. What do the coloured lines mean? Clutterness makes the Figure impossible to read. Maybe the mean values should be used instead.
Table 3: provide statistical analysis of the results presented in the table. Are the gender-related differences between top terms statistically meaningful?
Table 4: provide statistical analysis of the results presented in the table. Are the age-related differences between top terms statistically meaningful?
Figure 3: provide meaningful headers or explanations for columns 1-6. Why it is not presented as a table? Provide the statistical analysis of gender related differences of danger assessment results. Use ANOVA or similar method.
Table 7: the sum of images is short of 70000 mentioned in L. 253. What is the distribution of keywords from Table 6 between different risk categories?
Section 5 seems not related to other parts of the paper.
Section 6 should be improved and extended. Main results from survey must be used to support claims. Limitations of the methodology used should be discussed. Future research directions should be outlined.
300-303: remove entries for X.X.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article addresses an interesting issue, but it has many weaknesses and requires clarification:
Why is the research sample 250 participants? It is not known what is the statistical error of this sample and what level of significance was adopted?
Why elsewhere is the sample 154 respondents? Generally, a maximum sample error of + -5% (p <0.05) is assumed for such studies, but then the sample is 386 respondents. In this situation, sample 250 is too small.
The abbreviation avg (TF-IDF) does not explain. What does "??" mean?
Figure 1, I suggest to define as a table and improve the quality.
Figure 3 is not a drawing but a table (in addition of very poor quality) A more detailed description of the analysis methods used is necessary.
The Calinski Harabasz method comes from the names of two authors (Calinski-Harabasz method). Please complete the article with a description of this method of analysis (including the methodology of the index used).
The article has a very weak theoretical basis. It is necessary to increase the literature review and the number of scientific publications cited in the areas raised.
Chapters 5 and 6 should be seriously developed, as the article has neither proper discussion nor a summary.
The article is very messy. There are also many formal and graphic shortcomings. I recommend a thorough improvement.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have improved the paper significantly. All my concerns and suggestions were addressed properly. I recommend the paper to be accepted and published.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Dear Authors
It's much better now. Thank you for correcting the article.
I accept the changes made, however, please enter the numbering of the mathematical formulas (all) so that you can refer to them in the text of the article, e.g. (1).
Good luck!
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
