A Metric Learning Perspective on the Implicit Feedback-Based Recommendation Data Imbalance Problem

Round 1

Reviewer 1 Report

Technical Suggestions:

Provide more details on computation of context factors from text and citation data. A diagram showing this process would help.

Explain the motivations behind choosing the specific similarity fusion approach.

Clarify the differences between your negative sampling method and previous approaches.

More direct comparison of your approach to alternatives would strengthen the paper. Review and cite recent relevant papers in the final version:

1) P. Thantharate, "IntelligentMonitor: Empowering DevOps Environments with Advanced Monitoring and Observability," 2023 International Conference on Information Technology (ICIT), Amman, Jordan, 2023, pp. 800-805, doi: 10.1109/ICIT58056.2023.10226123.

2) Pagano, T.P.; Loureiro, R.B.; Lisboa, F.V.N.; Cruz, G.O.R.; Peixoto, R.M.; Guimarães, G.A.d.S.; Oliveira, E.L.S.; Winkler, I.; Nascimento, E.G.S. Context-Based Patterns in Machine Learning Bias and Fairness Metrics: A Sensitive Attributes-Based Approach. Big Data Cogn. Comput. 2023, 7, 27. https://doi.org/10.3390/bdcc7010027

Technical Questions:

Q1. How did you select the hyperparameters α and ω for the loss function? Some insight into how sensitive performance is based on these parameters would be useful.

Q2. For the context fusion method, did you experiment with other alternatives besides multiplication rule? If so, how did they compare?

Q3. The efficiency gains depend on precomputing the context factors. Would this be feasible for very large scale paper recommendation scenarios?

Q4. Did you consider any other contextual signals besides text and citation data? If not, what other signals could be incorporated in the future?

Author Response

Thank you very much for your kindly re-consideration and giving us a new opportunity to revise our work! According to your comments, we have tried best to modify our manuscript to meet with the requirements. In this revised version, changes to our manuscript within the document were all highlighted by using red colored text. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments for author File: Comments.pdf

Author Response

Thank you very much for your kindly re-consideration and giving us a new opportunity to revise our work! According to your comments, we have tried best to modify our manuscript to meet with the requirements. In this revised version, changes to our manuscript within the document were all highlighted by using red colored text. 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments:

The text is interesting and presents a new perspective for classifying large volumes of research articles. The article is well structured. The methodology is clear; the results are consistent and well discussed. I have a few comments:

1. remodelling the abstract so that it clearly presents the objective; methodology; results; practical implications and indications for future research;

2. The introduction is very brief and does not explicitly state the problem question or the objectives;

3. in Line 55, the word TOPIC is in upper case, is there a special reason? 

4. Do a little English revision. The titles of the figures don't all have end punctuation. Proofread the whole article.

All in all, good proofreading

Author Response

Thank you very much for your kindly re-consideration and giving us a new opportunity to revise our work! According to your comments, we have tried best to modify our manuscript to meet with the requirements. In this revised version, changes to our manuscript within the document were all highlighted by using red colored text. 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The manuscript was well revised following a set of my previous comments.