Improved DeepFM Recommendation Algorithm Incorporating Deep Feature Extraction
Round 1
Reviewer 1 Report
The authors describe a novel recommendation system that introduces a new feature fusion system as well as periodically retrained user preference features.
Overall, the authors did a good job. The existing literature review seems sound and the neural network was evaluated and compared in-depth with existing algorithms.
The authors should elaborate on their hyperparameter optimization approach. Only final (hyper-)parameters are reported.
The overall quality of the English somewhat diminishes the paper quality, because there are a lot of hard-to-understand sentences. Overall, the paper would drastically benefit from a more stringent and extensive argumentation with a clear common theme. Especially the first paragraph is particularly unconvincing.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper presents the interactive learning of low-order and high-order features. The following are the major concerns: -
1. What is the Deep FM? Please insert the abbreviation table for the sake of simplicity.
2. The literature review section needs more improvement and avoids long paragraphs.
3. In section 2.3, lines 236 and 244, The left & right sides of the model in figure 4, please correct this.
4. What is the YFM in equation 9, and how it can be calculated? Please elaborate in terms of mathematical expression.
5. The explanation of figure 4, needs a comprehensive explanation? Also, explain the Users and Items.
6. How many layers have been utilized in the evaluation of the DNN?
7. Dataset utilized to need more explanation? Simulation parameters are missing in the paper?
8. Why has the proposed algorithm been said many times improved Deep FM? Where is the improvement? Please improve the contribution section accordingly.
9. What is the computational complexity of the Improved Deep FM as compared to the existing algorithm?
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors proposed the recommendation system. Though, as the authors claim, the proposed scheme provides better results than some of the existing schemes. I recommend the following suggestions to improve the manuscript:
1. In the introduction section, some notable research work should be included.
2. For line number 102 to 114, the main contribution has been defined by the authors. Please revise it so that the actual contribution can be easily understood by the reader,
3. The reference order is wrong for [8] and [9].
4. The proposed scheme should clearly show the authors' contribution.
Author Response
Point 1: In the introduction section, some notable research work should be included.
Response 1: The literature review section has been removed the overview section and some notable research works have been added (in section 1, line36-line71).
Point 2: For line number 102 to 114, the main contribution has been defined by the authors. Please revise it so that the actual contribution can be easily understood by the reader.
Response 2: The main contribution has been redefined (in section 1, line79-line87).
Point 3: The reference order is wrong for [8] and [9].
Response 3: References have been revised
Point 4: The proposed scheme should clearly show the authors' contribution.
Response 4: The authors' contribution has been revised (in section 1, line79-line87).
Author Response File: Author Response.pdf
