Study on the Technology Trend Screening Framework Using Unsupervised Learning
Round 1
Reviewer 1 Report
1. line 60. “term frequency-inverse term frequency” (TF-IDF) do you mean term frequency-inverse document frequency.
2. how to define noise data and outliers? Why noisy data can’t be outliers?
3. line 356, we are voting for the outliers. “In this study, when an outlier is selected from two or more methods, it is judged as the final outlier description document.” For me it is more like hard voting, do we consider soft voting?
4. Line 372, “In this respect, because the proposed method classifies the recently 372 applied patents as outliers, this method is capable of screening technology trends.” Do you mean the new technology will always be an outlier? Is it possible that only one or two topics are very popular for a short time and no longer popular since then but it could be classified as outlier? I guess I don't quite understand the practical application of the method.
5. Please proofread, there are some typo. For example: line 320, 18 months not “18 monts” and “decrease” not “decrese”
6. Line 320 “Since it takes 320 18 monts for a patent to be published, we do not consider the drastic decrese in drone- 321 related patents after 2018 in our research.” What does “not consider”? I saw you still use the data till 2019.
7. How to evaluate the model accuracy?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
The paper presents a good overall approach to recognizing patterns in documents using the proposed methodology using the doc2vec model.
The paper is well-written and easy to follow.
In my opinion, it is good to be published.
Just some suggestions for the authors in future work:
One can also use One-Class learning to analyze the normal data to detect novelty or anomalies. Deep Autoencoders can also be helpful for future work, where XAI techniques can be used to report why a document is good or anomalous.
Another suggestion is table 8 could begin on the next page. The header got separated from the rest of the content.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
· The contributions are not clear, I suggest to authors to write in good manner to be attractive to readers on introduction section
· State of the art need be addressed to clarify the main contributions in this document
· I suggest to authors to do a hole review about write style to verify grammars mistake
Comments for author File: Comments.docx
Author Response
Please see the attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
All comments are addressed
Author Response
We are grateful to the reviewer for the critical comments and useful suggestions that have helped us to improve our paper considerably.
Thank you.