Which Influencers Can Maximize PCR of E-Commerce?
Round 1
Reviewer 1 Report
A brief summary
This paper explored the effectiveness of influencer marketing on e-commerce and provided valuable insights to help businesses improve their online sales. The study found that by selecting the right influencers, purchase conversion rates could be significantly increased. The paper also introduced a multi-modal micro influencer analysis scheme that utilized data from Mecha Solution's own shopping mall and famous Korean internet platforms such as Coupang and Naver.
Specific comments:
This paper has a clear motivation.
I am leaning to give a "major revisions" based on my current knowledge and understanding of the paper. But I will be willing to revisit the decision after get feedback from the author(s).
In particular, I would be glad if the author could clarify the questions below.
*The paper does not provide a comprehensive analysis of the limitations and challenges of influencer marketing in e-commerce.
*The study only focuses on the Korean market, which limits the generalizability of the findings to other regions and cultures.
*The paper does not address ethical concerns related to influencer marketing, such as transparency and authenticity.
*The sample size used in the study may not be representative of the entire population, which could affect the validity of the results.
*The paper does not provide a detailed explanation of how the multi-modal micro influencer analysis scheme works, which could make it difficult for readers to replicate or apply it in their own research.
*The deep learning methods used in the paper are very basic methods and do not go for SOTA methods such as Transformer based various methods, etc.
*A paragraph should be dedicated to summarize the contribution and novelty of the paper.
*Many of the figures and tables appear in Korean, so English translations are required.
*Many of the figures are blurry and need to be described in vector graphics.
The English does not read very smoothly and needs further embellishment.
Author Response
First of all, thank you for your specific review. I replied to your review and attached my manuscript reflecting it. I haven't corrected my English yet, but please focus on the contents
*The paper does not provide a comprehensive analysis of the limitations and challenges of influencer marketing in e-commerce.
Response: I added limitations of that
*The study only focuses on the Korean market, which limits the generalizability of the findings to other regions and cultures.
Response: I added that limitations
*The paper does not address ethical concerns related to influencer marketing, such as transparency and authenticity.
Response: Our data sources are various. Data with less anonymity was masked and most of the data is publicly available and anonymous.
*The sample size used in the study may not be representative of the entire population, which could affect the validity of the results.
Response: I added that limitations
*The paper does not provide a detailed explanation of how the multi-modal micro influencer analysis scheme works, which could make it difficult for readers to replicate or apply it in their own research.
Response: I added some figure that shows scheme works and add some description
*The deep learning methods used in the paper are very basic methods and do not go for SOTA methods such as Transformer based various methods, etc.
*A paragraph should be dedicated to summarize the contribution and novelty of the paper.
Response: I added contribution
*Many of the figures and tables appear in Korean, so English translations are required.
Response: I added some translations on figures. However, as WordCloud figures, I just kept Korean. Instead, the text was modified by adding parentheses and English translations next to Korean.
*Many of the figures are blurry and need to be described in vector graphics.
Response: I modified about it
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This study focuses on leveraging the web as an e-commerce medium and analyzing recommender systems to estimate customer interest. It introduces a multi-modal micro influencer analysis scheme that utilizes deep learning techniques and diverse data sources, including article postings, comments, and statistics from popular Korean internet platforms. The scheme aims to predict the purchase conversion rate of influencers and recommends them using content-based collaborative filtering and user-based collaborative filtering. The proposed scheme is implemented and experimentally validated to demonstrate its effectiveness in achieving marketing maximization goals.
The manuscript is interesting but in the current state it cannot be accepted for publication. There are several issues that the authors could consider addressing in order to improve its quality. These, along with some comments and suggestions, are given in the following:
- Related Work could be enhanced. In particular, in the current state, the section is long and not very informative. The authors could try removing less related studied and add ones that are more related and recent.
- As for related work, the authors could consider citing the following works which are extremely close to the proposed one [doi.org/10.1016/j.datak.2022.101979, doi.org/10.3390/bdcc6040130].
- A specific definition of “influencer” should be provided.
- The introduction could report the outline of the paper, i.e., a description of sections and their content.
- Data reported in Table 1 seem to be inconsistent. For instance, the value for Image num on the first column is equal to the value for Sympathy num on the second column. Please check it.
- A workflow of the proposed approach is needed. I suggest the authors to add a Section whose aim is to depict the workflow clearly. The authors could also use a graphical depiction of it.
- Word clouds are really not informative in the current version. The authors could consider enhancing them with English ones. The same holds for the pie charts in Figures 23 and 24.
- Figure 28 should be reported as a table.
- What are the limitations of the proposed work?
- Conclusion are needed as well as a brief discussion of future works.
Author Response
First of all, thank you for your specific review. I replied to your review and attached my manuscript reflecting it. I haven't corrected my English yet, but please focus on the contents
- Related Work could be enhanced. In particular, in the current state, the section is long and not very informative. The authors could try removing less related studied and add ones that are more related and recent.
- As for related work, the authors could consider citing the following works which are extremely close to the proposed one [doi.org/10.1016/j.datak.2022.101979, doi.org/10.3390/bdcc6040130].
Response: I revised it considering the corresponding reply.
- A specific definition of “influencer” should be provided.
Response: I revised it considering the corresponding reply.
- The introduction could report the outline of the paper, i.e., a description of sections and their content.
Response: I revised it considering the corresponding reply.
- Data reported in Table 1 seem to be inconsistent. For instance, the value for Image num on the first column is equal to the value for Sympathy num on the second column.
Response: It was confirmed that miscommunication occurred while receiving information and was entered incorrectly. This is my mistake. The corresponding content has been modified.
- A workflow of the proposed approach is needed. I suggest the authors to add a Section whose aim is to depict the workflow clearly. The authors could also use a graphical depiction of it.
Response: I revised it considering the corresponding reply.
- Word clouds are really not informative in the current version. The authors could consider enhancing them with English ones. The same holds for the pie charts in Figures 23 and 24.
Response: I added some translations on figures. However, as WordCloud figures, I just kept Korean. Instead, the text was modified by adding parentheses and English translations next to Korean.
- Figure 28 should be reported as a table.
Response: I revised it considering the corresponding reply. However, I thought input was important, so I added it together.
- What are the limitations of the proposed work?
Response: I revised it considering the corresponding reply.
- Conclusion are needed as well as a brief discussion of future works.
Response: I revised it considering the corresponding reply.
Author Response File: Author Response.pdf
Reviewer 3 Report
Overall, the paper presents a comprehensive analysis scheme for maximizing the purchase conversion rate in e-commerce marketing through micro influencers. The proposed scheme leverages multi-modal data, including real user article postings and comments, to assign influencer scores and provide recommendations for items that have not yet been reviewed. The implementation and experimentation of the scheme demonstrate its successful achievement of the desired goal.
However, there are several major problems that the authors should address in their review:
-
In the Introduction section, the specific research gap that this paper aims to address is not clearly established. The authors should articulate the novel contribution of their work and how it differs from existing studies. Additionally, they should provide a stronger justification for the proposed approach and explain its potential benefits and implications.
-
The Related Works section lacks a comprehensive synthesis and critical evaluation of existing research. The authors should analyze the strengths and weaknesses of previous studies, identify gaps in the literature, and clearly articulate how their proposed approach fills those gaps. They should also provide a stronger connection between previous works and their own study.
-
The Background section fails to establish the direct relevance of concepts like RNNs, LSTMs, TF-IDF, and collaborative filtering to the research topic. The authors should clearly explain why these concepts are necessary and how they specifically contribute to the proposed scheme. More detailed and comprehensive explanations, supported by references or citations, would enhance reader understanding.
-
The Research Method section lacks detailed explanations and justifications for the specific methodologies used. The authors should provide more clarity on the tools, algorithms, and techniques employed, and explain how they contribute to addressing the research objectives. Additionally, they should discuss data quality and limitations, including potential biases and their impact on the research outcomes.
-
The Proposed Scheme section requires more clarity and justification for the chosen scoring metrics and their respective weights. The authors should explain why specific features and ratios were selected and how they contribute to measuring the quality and influence of influencer posts. They should also provide a more detailed explanation of the influencer recommender system, including the algorithms and methods employed.
-
The Results section lacks empirical evidence and validation. The authors should present results from experiments or case studies that demonstrate the effectiveness and accuracy of the designed system in real-world scenarios. They should also discuss limitations and challenges, such as assessing cumulative influence and addressing influencers with lower visitor numbers but high influence.
Furthermore, it is recommended that the authors cite the following papers to strengthen their study on related literature: doi.org/10.3390/bdcc6040130 and doi.org/10.1016/j.datak.2022.102048. This will provide additional support and context to their research and demonstrate a comprehensive understanding of the existing body of knowledge.
Overall, with these improvements and additions, the paper will significantly enhance its contribution to the field of influencer marketing and analysis.
The English quality of the paper is generally good, but there are instances where sentence structure could be improved for better clarity and readability. Some sentences are overly long and complex, and there are a few grammatical errors and inconsistencies in punctuation that could be addressed through careful proofreading.
Author Response
First of all, thank you for your specific review. I read your review and attached my manuscript reflecting it. I haven't corrected my English yet, but please focus on the contents
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors answered most of the issues. But still does not explain why the Transformer-based language model was not used.
The English translation of this paper needs further editing.
Author Response
I revised the manuscript reflecting your feedback and completed the English revision.
Author Response File: Author Response.pdf
Reviewer 2 Report
The authors only partially addressed my concerns. In particular, the authors reported that the suggested references were cited. However, this claim does not constitute the truth. The authors could consider citing the following related work: [doi.org/10.1016/j.datak.2022.101979]
Author Response
I revised the manuscript reflecting your feedback and completed the English revision. I used paper that you recommend on related works section. Thank you for feedback.
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors have successfully answered to all my questions.
Author Response
I completed the English revision. I would appreciate it if you could tell me if there is anything else I need to fix.
Author Response File: Author Response.pdf
