A New Method for Graph-Based Representation of Text in Natural Language Processing
Round 1
Reviewer 1 Report
The manuscript analyses a text representation method based on a new graph representation to present its application to book classification and examines the possibility of using the proposed method of text representation, which in the next steps will be used to train machine learning models in order to classify them.
However, the structure and method of the manuscript are not clear, greatly increasing the difficulty of reading. And the manuscript has the following problems that need to address:
1. The abstract did not express the proposed method clearly, making it difficult for readers to clarify the main idea of the manuscript.
2. A framework figure is required to demonstrate the proposed method.
3. The proposed method also needs the Algorithm to illustrate.
4. The author needs to clearly introduce the settings used in the experiment.
5. The author needs to add analyses of the experimental results for each table.
The grammar of the manuscript needs to be comprehensively corrected. For example, in the Abstract, “The aim of this paper is to analyse a text representation”-> The aim of this paper is to analyze a text; “allowed to achieve accuracy, precision, recall and F1–score at the level of over 90%.”-> allowed us to achieve accuracy, precision, recall, and an F1–score at a level of over 90%.……
Author Response
All responses for remarks are attached as a file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Authors ,
This paper aims to propose a novel method for text representation. The idea of this paper is clearly defined however the information flow and writing style is difficult for readers. In order to improve the paper's quality, the authors are required to consider the following comments.
1- Words mismatch, in the paper you use words such as “the aim of this paper “ , “this work “, “ this article “, and “ this research “. Please be consistent.
2- List the paper contributions.
3- Draw the research methodology in the figure that will be the reader a full picture.
4- Elaborate more with examples of how you represent the words in the form of a graph.
5- In this paper, the contribution and methods should be clearly explained in the methodology section.
The paper requires minor proofreading.
Author Response
All responses for remarks are attached as a file.
Author Response File: Author Response.pdf
Reviewer 3 Report
The results of the proposed approach seem to be promising and the paper is clearly written. The two main issues are related to the dataset used and to the word representation techniques employed. The authors should evaluate their proposed representation technique also on other datasets and also comparing it to modern representation techniques based on word embedding. Some of the comments that I think should be addressed by the authors are reported below.
Line 156: Lametization instead of Lemmatization
Line 186: Space character is missing after Gutenberg.
Line 244: "natural language processing" is used. But the acronym NLP was already introducted.
Line 260: "lametization" again...
Table 3-6: please highlight the best results
Section 3.3: Some references to text representation for attention based models are missing. Please include a discussion on word embedding.
Section 3.5: Novel state-of-the-art approaches proposed at popular international shared tasks for text classification are missing. They are based on ensemble, CNN and Hybrid LSTM and have proved to be effective as well as Transformers. So please add some more recent and relevant literature to the related works. Some of them are:
1) An SVM Ensamble Approach to Detect Irony and Stereotype Spreaders on Twitter, Croce et al., 2022
2) NLNDE at SemEval-2023 Task 12: Adaptive Pretraining and Source Language Selection for Low-Resource Multilingual Sentiment Analysis, Wang et al., 2023
3) Improving Irony and Stereotype Spreaders Detection using Data Augmentation and Convolutional Neural Network, Mangione et al., 2022
4) SemEval-2022 Task 6: iSarcasmEval, Intended Sarcasm Detection in English and Arabic, Farha et al., 2022
5) An Ensemble Model Using N-grams and Statistical Features to Identify Fake News Spreaders on Twitter., Buda et al., 2020
The paper is well written and needs a further reading for minor mistakes correction.
Author Response
All responses for remarks are attached as a file.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
(1) please refine the contributions of this paper to make it more understandable. (2) please consider if it is possible to combine table 1 and table 2 to a single table. (3) Figure 1 is not good enough. What Id means?
please polish the paper carefully.
Author Response
All responses for remarks are attached as a file.
Author Response File: Author Response.pdf
Reviewer 2 Report
Well done.
Author Response
Thank you very much -- we have tried to improve our article according to all the comments. We are glad that we met your expectations.
Reviewer 3 Report
Please double check any mistake in the references format. Sometimes it is written. For example, references 60 and 62 report "et al." but no other authors are included. Please double check author names, titles and venues. Eventually using Scopus.
Minor mistakes.
Author Response
All responses for remarks are attached as a file.
Author Response File: Author Response.pdf
