All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
can't comment because I can't see previous issues
can't comment because I can't see previous issues
can't comment because I can't see previous issues
please write a proper rebuttal
We noted that in some parts your text should be improved.
As we don’t offer language editing when articles are in production, it is definitely recommended that you use an editing service to check the language.
[#PeerJ staff note: Although the Academic Editor is happy to accept your article as being scientifically sound, a final check of the manuscript shows that it still needs another round of English editing. Please can we ask you to edit the language one last time, as we do not perform language editing as part of our production process#]
no comment
no comment
no comment
All comments have been addressed.
no comments
no comments
no comments
All the points raised by the reviewers were addressed by the authors. Now the text and figures are clearer and ready for publication.
The authors have addressed several issues, however the referees pointed out that some issues still remain. Please address all referees' comments.
All my comments have been addressed. However, the compilation of some references should be double-checked, e.g., see "Opinion Miningwith"
See my previous review
See my previous review
All the comments that were suggested in the previous revision have been addressed. However the captions of Figure 4 and some Tables presents some errors that need to be reviewed:
Figure 4: F1score
Table 1: Rrepresent
Table 2: Exampleof worddistribution
Table 3: interms
Table 4: Comparisonresults
no comment
no comment
The revisions are satisfactory, except that some minor errors need to be reviewed (see basic reporting).
no comments
no comments
no comments
The manuscript was greatly improved in this revised version. The main points raised by the reviewers were addressed by the authors. However, I have some minor comments in order to make the manuscript clearer:
The names of the datatsets do not need to be mentioned in the abstract (line 32) unless they would be recognized just by their names.
The second paragraph of section "Related Work" of the tracked changes version is different from the final revised version. The same thing happend in Section 3.1. The sentence starting with "Similar to [27] each word in the sentence... " in the revised version is not present in the final version. The authors should check if both versions are the same.
Line 94: a reference is missing for the SemEval challenge.
Line 114 should be corrected as "For instance, 'Author et al.' [17] employed RMB....". The authors should check other places in the manuscript with similar notation, mainly in the Related Work section. It is also weird to start a sentence with "[n]" as the example of line 115. I recommend using the name of the authors instead or rearranging the sentence.
Still, some figures are imcomplete. In Figs. 1, 2 and 3, the y-axis title is missing. It should refer to performance. In addition, in Fig. 3, the x-axis is not there.
Line 147: "...is made up of..."
Caption of Table 2: "word distribution"
Caption of Table 4: "Comparison results"
Line 306: "... baseline models"
Please address all the referees' comments.
see below
see below
see below
The manuscript is centered on an interesting topic.
Organization of the paper is good and the proposed method is quite novel.
The manuscript, however, does not link well with recent literature on sentiment analysis appeared in relevant top-tier journals, e.g., the IEEE Intelligent Systems department on "Affective Computing and Sentiment Analysis". Also, check latest trends in deep learning based aspect extraction, e.g., attentive LSTM.
Many issues with acronyms: some acronyms are used without being defined first, e.g., CNN, some other are defined but then never used, e.g., CRF.
Some bad English constructions and misuse of articles: a professional language editing service is strongly recommended (e.g., the ones recommended by IEEE, Elsevier, and Springer) to sufficiently improve the paper's presentation quality for meeting PeerJ's high standards.
• The article is generally written in a clear, coherent and professional English. However, there are some segments that need to be corrected because they are not well understood or they are incorrectly written. For the next revision, it is necessary to improve the segments at lines: 13, 38, 40, 88 to 91, 92, 95 to 97, 153, 191 to 192, 262, 320 to 321, 396 to 397, and 476.
• The Introduction and background are correct because they reveal the problems that need to be addressed and clearly explain the objectives of the paper. And the related works are generally well-founded because they explain clearly the works that have addressed the problem of aspect extraction.
• The work is very well structured.
• The tables and figures are relevant and help to better understand the methodology and results obtained, however, figures 2 and 3 need to be edited and eliminate the title left by mistake in both figures: “Chart Title”. In addition, it would be good to explain a little better all the figures in the caption that corresponds to them.
• Conclusion of basic reporting: Approved
• The research of the article satisfies the Aims and Scope of the journal
• The problems existing with the aspect extraction of reviews task and how these problems can be solved with the paper proposal are clearly specified. In addition, the suggested methods are quite relevant and novel for the current literature. As a suggestion, the research question should be clearly specified so that it can then be answered in the conclusions of the work.
• Many works found in the literature have been explained, which is quite good. However, it would be ideal to explain a little more the advantages and disadvantages of those works that you consider most important and that are better related to the proposal of your work, that is, you might detail a little more explaining their performance.
• The methodology of the work is clearly explained, the methods are described in detail.
• In the part of the explanation of the datasets, it would be interesting to show some examples of the most important aspects that are found in each dataset, no example was shown. Explain a little better the datasets used.
• Conclusion of experimental design: Approved
• The article showed a large number of experiments with relevant results. The data is robust and statistically sound and meets the objectives that were raised in the introduction of the article.
• The conclusions are well established and explain the importance of the results that have been obtained. However, since the research question has not been clearly defined previously (in the introduction), the conclusion seems a bit incomplete. The research question should be defined and finally answer that question according to the results obtained in this work.
• Conclusion of validity of the findings: Approved
The manuscript entitled "Aspect extraction on user textual reviews using Multi-Channel Convolutional Neural Network" presents an approach for aspect extraction using deep multichannel convolutional neural network. The main topic is relevant and interesting to the scope of this journal. I would like to highlight some points about the general organization of the paper that should be revised by the authors.
- I recommend that the manuscript be reviewed for what concerns the English language for both the composition of the sentences and the grammar.
- The text also needs more attention with some sentences. For instance, "As a feature direction of the research" (line 480)
- There are many abbreviations in the introduction that were not defined in the text, like "LSTM".
- "CRF" is defined twice.
- Some figures need to be improved. In Fig. 2 it is missing the legend for the orange bar. In Fig. 3, the x-axis is not shown.
The literature references and the related works are well structured and the main background of the manuscript is provided.
The main theoretical tools approached by the authors are cutting-edge research on many machine learning and related topics nowadays and the authors explore their architecture for aspect extraction within the context of sentiment analysis. I missed some motivation for what concerns this main application of the method. I think that the authors could improve the text by adding some real-world examples of the datasets that were used. This way, they can illustrate the main application of the proposed approach and motivate the reader to go further in the methodology and the results.
The research question is well defined in the manuscript. The authors performed a deep investigation of the results by making comparisons with other methods in literature and also with some variations of the proposed model. The authors showed that the proposed approach could outperform many state of the art methods. Therefore, the main contribution of the manuscript is the performance gain obtained by the proposed method.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.