An Improved African Vulture Optimization Algorithm for Feature Selection Problems and Its Application of Sentiment Analysis on Movie Reviews

Round 1

Reviewer 1 Report

The author has presented a comprehensive improvised AVOA with Hyper-heuristic and Multi-strategy for Feature Selection Problems. The manuscript also presented a case study using the proposed algorithm for Sentiment Analysis on Movie Reviews.

The manuscript is well written and analysed in depth. The novelty for feature selections has been done with wide scope consideration. There are various algorithms developed and tested to determine the best algorithm. In addition, the proposed algorithm setting and parametric values have been well defined for future researchers. The proposed model has been validated with a few recently published algorithms with analysed data from a common database.

Furthermore, case study analyses for movie review application is presented to further understand the algorithm capability. References are relevant and recent. Results are plotted well. Data is well tabled. Sufficient discussion made. Thus, I recommend this manuscript for publication.

Author Response

Manuscript ID: BDCC-1854586

Type of manuscript: Article

Title:   An Improved African Vulture Optimization Algorithm  for Feature Selection Problems and Its Application of Sentiment Analysis on Movie Reviews

BDCC

Dear Reviewer#1

Thank you very much for reviewing our manuscript. We also greatly appreciate the reviewers for their complimentary comments and suggestions. We have carried out the works that the reviewers suggested and edited the manuscript accordingly.

Reviewer #1:

The author has presented a comprehensive improvised AVOA with Hyper-heuristic and Multi-strategy for Feature Selection Problems. The manuscript also presented a case study using the proposed algorithm for Sentiment Analysis on Movie Reviews.

The manuscript is well written and analysed in depth. The novelty for feature selections has been done with wide scope consideration. There are various algorithms developed and tested to determine the best algorithm. In addition, the proposed algorithm setting and parametric values have been well defined for future researchers. The proposed model has been validated with a few recently published algorithms with analysed data from a common database.

Furthermore, case study analyses for movie review application is presented to further understand the algorithm capability. References are relevant and recent. Results are plotted well. Data is well tabled. Sufficient discussion made. Thus, I recommend this manuscript for publication.

Response: I express my utmost appreciation and thanks for the statements of the honorable reviewer who have allocated their valuable time to review this paper.

Finally, we would like to express our gratitude for the efforts of the editor-in-chief and the esteemed secretaries and reviewers of this journal. We hope that their efforts will be more helpful in developing this journal. We are very grateful for the help they have given us to improve our paper. Be proud and successful.

With my best and warm regards

Farhad Soleimanian Gharehchopogh, P.hd

Corresponding Author

Reviewer 2 Report

The idea of the article is interesting indeed, however, there are some points that must be addressed by the authors:

1. Problem statement in missing in the Abstract and Introduction. Authors should clearly explain what kind of problems are solved by the study and in what sphere?

2. Related work section is too small, authors have mentioned only few literature review that is not enough to justify the statements of the authors regarding disadvantages of the study. For example: line no 180. "Inefficiency of algorithms in high dimentional datasets". Do you have any reference or citation who claimed this statement? In my opinion, there are various approaches and studies which have been done on high dimensional datsets used in this study and they provided very good results.

Similarly for all other disadvantages that authors mentioned in the related work section. Also, for the last, "evaluation of algorithms on few datasets". The dataset used in any study must be relavent to the study and it must have a good number of samples to train and test the model and such studies are a lot in scientific literature. If authors are talking about using several diffrent type of datasets to be used in the study then in ML/Data Mining, such things are usually not considered because all datas sets have different kind of attributes, features and importance. Therefore, it is not feasible to develop one Universal model to classifiy and predict medical diseases and flower properties together.

3. In this study, the authors have used various different datasets, which are related to medical diseases, wine, flowers, steel, soyabean. And they compared the results with some techniques.

4. If authors really claims that there algorithm is able to outperform on previous studies on every dataset then authors must compare their approach with best studies so far done on every datasets. Then only, their claim will be justfied that their approach is a universal approach to handle all kind of datasets used in this study.

Author Response

Manuscript ID: BDCC-1854586

Type of manuscript: Article

Title:   An Improved African Vulture Optimization Algorithm  for Feature Selection Problems and Its Application of Sentiment Analysis on Movie Reviews

BDCC

Dear Reviewer#2

Thank you very much for reviewing our manuscript. We also greatly appreciate the reviewers for their complimentary comments and suggestions. We have carried out the works that the reviewers suggested and edited the manuscript accordingly.

Reviewer #2:

The idea of the article is interesting indeed, however, there are some points that must be addressed by the authors:

1. 1. Problem statement in missing in the Abstract and Introduction. Authors should clearly explain what kind of problems are solved by the study and in what sphere?

Response: Thanks to the opinion of the respected reviewer, the abstract and introduction sections have been edited, and I hope the respected reviewer will accept it.

2. Related work section is too small, authors have mentioned only few literature review that is not enough to justify the statements of the authors regarding disadvantages of the study. For example: line no 180. "Inefficiency of algorithms in high dimentional datasets". Do you have any reference or citation who claimed this statement? In my opinion, there are various approaches and studies which have been done on high dimensional datsets used in this study and they provided very good results. Similarly for all other disadvantages that authors mentioned in the related work section. Also, for the last, "evaluation of algorithms on few datasets". The dataset used in any study must be relavent to the study and it must have a good number of samples to train and test the model and such studies are a lot in scientific literature. If authors are talking about using several diffrent type of datasets to be used in the study then in ML/Data Mining, such things are usually not considered because all datas sets have different kind of attributes, features and importance. Therefore, it is not feasible to develop one Universal model to classifiy and predict medical diseases and flower properties together.

Response: Thanks to the respected reviewer's opinion, we have improved the previous work section, and I hope the reviewer will accept it. In the meantime, we modified the entire previous work. And we fixed the existing problems. The sentence in line 180 was a mistake on our part. We deleted it, and the honorable judge also mentioned: )Therefore, it is not feasible to develop one universal model to classifiy and predict medical diseases and flower properties together).  This statement is accurate, and we cannot build a model that will work for all datasets. By all the datasets, we meant the datasets in this paper. And our generalization of all datasets was a mistake. We have corrected this sentence. And we express our utmost appreciation and thanks to the respected reviewer for this insight, which improved our paper.

3. In this study, the authors have used various different datasets, which are related to medical diseases, wine, flowers, steel, soyabean. And they compared the results with some techniques. If authors really claims that there algorithm is able to outperform on previous studies on every dataset then authors must compare their approach with best studies so far done on every datasets. Then only, their claim will be justfied that their approach is a universal approach to handle all kind of datasets used in this study.

Response:  While appreciating and thanking the esteemed reviewer and thanking him for his insight, we have checked and corrected the whole article on this matter that they said. Our sentence or sentence, in this case, was interpreted ambiguously, so we fixed this. It means all the datasets, proposed and comparative algorithms, which are mentioned in the text of this paper . It is worth mentioning that in this article, the datasets (abalone, breastcancerw, tictactoe, glass, heart, wine, letterre recognitions, seismicbumps, spect, german, waveform, breastEW, Steel, Dermatology, ionosphere, soybean, lungcancer, spambase sonar , libras, audiology, krvskp, LSVT, PersonGaitDataSet, pd_speech, ORL, warppie, lung, SMK-CAN-187) and optimization algorithms (BBA, BDA, BCCSA, BFFAG, BGWO, BPSOT) have been used and evaluated. We mean these algorithms and datasets, not all existing algorithms and datasets.

Finally, we would like to express our gratitude for the efforts of the editor-in-chief and the esteemed secretaries and reviewers of this journal. We hope that their efforts will be more helpful in developing this journal. We are very grateful for the help they have given us to improve our paper. Be proud and successful.

With my best and warm regards

Farhad Soleimanian Gharehchopogh, P.hd

Corresponding Author

Reviewer 3 Report

This paper provides two versions based on the s-shaped and v-shaped transfer functions of African Vulture Optimization Algorithm and Binary African Vulture Optimization Algorithm with Hyper-heuristic.

In my opinion, this paper needs some improvements before publication.

In particular, the overall presentation needs improvement and the manuscript should be reorganized.

First of all, the introduction section should be reorganized by better discussing the addressed problem.

The discussion on related works should be improved. The following works are suggested as a starting point for this discussion:

 ·         de Carvalho, V.R.; Özcan, E.; Sichman, J.S. Comparative Analysis of Selection Hyper-Heuristics for Real-World Multi-Objective Optimization Problems. Appl. Sci. 2021, 11, 9153. https://doi.org/10.3390/app11199153

·         Abiodun, E.O., Alabdulatif, A., Abiodun, O.I. et al. A systematic review of emerging feature selection optimization methods for optimal text classification: the present state and prospective opportunities. Neural Comput & Applic 33, 15091–15118 (2021). https://doi.org/10.1007/s00521-021-06406-8

·         Montazeri, Mitra. ‘HHFS: Hyper-heuristic Feature Selection’. Intelligent Data Analysis 20(4):953-974.  2016. DOI: 10.3233/IDA-160840

The authors should provide a discussion characterizing each work, its benefits and drawbacks, and a comparison among the related works to justify the necessity of the study proposed in the paper. Please state clearly and precisely in the paper what makes this work original.

Finally, there are typos (e.g. line 812: 79à0.79)

Author Response

Manuscript ID: BDCC-1854586

Type of manuscript: Article

Title:   An Improved African Vulture Optimization Algorithm  for Feature Selection Problems and Its Application of Sentiment Analysis on Movie Reviews

BDCC

Dear Reviewer#3,

Thank you very much for reviewing our manuscript. We also greatly appreciate the reviewers for their complimentary comments and suggestions. We have carried out the works that the reviewers suggested and edited the manuscript accordingly.

Reviewer #3:

1. This paper provides two versions based on the s-shaped and v-shaped transfer functions of African Vulture Optimization Algorithm and Binary African Vulture Optimization Algorithm with Hyper-heuristic. In my opinion, this paper needs some improvements before publication.

In particular, the overall presentation needs improvement and the manuscript should be reorganized. First of all, the introduction section should be reorganized by better discussing the addressed problem.

Response: I express my utmost appreciation and thanks for the honorable reviewer statements. The things mentioned in the text of the article have been corrected, I hope the respected referee has liked it.

2.The discussion on related works should be improved. The following works are suggested as a starting point for this discussion:

• de Carvalho, V.R.; Özcan, E.; Sichman, J.S. Comparative Analysis of Selection Hyper-Heuristics for Real-World Multi-Objective Optimization Problems. Appl. Sci. 2021, 11, 9153. https://doi.org/10.3390/app11199153

• Abiodun, E.O., Alabdulatif, A., Abiodun, O.I. et al. A systematic review of emerging feature selection optimization methods for optimal text classification: the present state and prospective opportunities. Neural Comput & Applic 33, 15091–15118 (2021). https://doi.org/10.1007/s00521-021-06406-8

• Montazeri, Mitra. ‘HHFS: Hyper-heuristic Feature Selection’. Intelligent Data Analysis 20(4):953-974. DOI: 10.3233/IDA-160840

The authors should provide a discussion characterizing each work, its benefits and drawbacks, and a comparison among the related works to justify the necessity of the study proposed in the paper. Please state clearly and precisely in the paper what makes this work original.

Response: Thanks to the respected referee who sent us good research, we read and used these articles in our paper.

3. Finally, there are typos (e.g. line 812: 79à0.79)

Response: I express my utmost appreciation and thanks for the honorable reviewer statements. The things mentioned in the text of the article have been corrected, I hope the respected referee has liked it.

Finally, we would like to express our gratitude for the efforts of the editor-in-chief and the esteemed secretaries and reviewers of this journal. We hope that their efforts will be more helpful in developing this journal. We are very grateful for the help they have given us to improve our paper. Be proud and successful.

With my best and warm regards

Farhad Soleimanian Gharehchopogh, P.hd

Corresponding Author

Round 2

Reviewer 2 Report

Accept as authors revised the article