New Hybrid Techniques for Business Recommender Systems
Round 1
Reviewer 1 Report
The authors proposed a weight-based hybrid strategy for business recommendations using existing recommendation techniques. However, the authors did not discuss the data processing pipeline, particularly, the pre-processing steps are not clear. I have the following suggestions for authors (The comments are not in the order of importance).
- Summarize the related works in the form of a table.
- Discuss the data preparation steps in detail. For instance, how the vectorization of input data was achieved?
- A detailed analysis of the results obtained has to be presented.
- What is the computational complexity of the proposed hybrid technique?
Author Response
Thank you very much for your feedback. We have worked on it and included the explanations in the attached response. All changes in the manuscript have been highlighted in the revised pdf version.
Best regards,
Authors
Author Response File: 
Author Response.docx
Reviewer 2 Report
Nice presentation on the business recommender system
Author Response
Thank you very much for your feedback!
Best regards,
Authors
Reviewer 3 Report
The article describes a hybrid recommender system for business consultancy, based on case-based reasoning and graph-based recommendation. It shows that the weighted hybrid approach outperforms the individual recommenders at all levels of query verbosity. The paper is clear, relevant for the field and written in a well-structured manner. It is also scientifically sound, follows a coherent methodology and presents very promising results. However, some critical decisions concerning the design of the solution are not well justified and rely on rather outdated cited publications. For example, regarding the suitability of CBR in business recommenders. Why not considering model-based reasoning as well? In general, most of the references are not current and thus should be updated. I would suggest the authors to perform a State-of-the-Art analysis, in order to support the proposed architecture and highlight the novelty of the research outcomes.
Author Response
Thank you very much for your feedback. We have worked on it and provided our response in the attached document. All changes in the manuscript have been highlighted in the revised pdf version.
Best regards,
Authors
Author Response File: Author Response.docx
Reviewer 4 Report
This article is devoted to the development of new methods for recommender systems. It is hard to deny that this is a very hot topic of research and hybrid methods are the most promising. The authors are developing a hybrid design of recommender systems. Their experiments show that this improves efficiency.
My comments:
- Table 1 shows the importance weights of various attributes. These weights are derived from the processing of consultants' opinions. So, in my opinion, the authors should point at indicators of consistency and variability of consultants' opinions.
- As an indicator of the effectiveness of the recommender system, the authors use MAP (mean average precision over the recommendation sets). However, there are other indicators of information search efficiency. This, for example, is the metrics of diversity, novelty, coverage, etc. I would like to see the comments of the authors regarding various performance metrics.
- In my point of view, the dependence in Figure 3 is best done in 3D. This will show the change in the alpha parameter from different arguments.
- For all experimental results, in my opinion, accuracy and reliability should be indicated.
Author Response
Thank you very much for your feedback. We have worked on it and included our response in the attached document. All changes in the manuscript have been highlighted in the revised pdf version.
Best regards,
Authors
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors addressed all the points awarded in the previous review round. Hence, it is recommended for publication.
Reviewer 3 Report
The authors have addressed all my comments in a satisfactory way.
