Biomedical Text NER Tagging Tool with Web Interface for Generating BERT-Based Fine-Tuning Dataset
Round 1
Reviewer 1 Report
It is unclear what the authors' original contributions are and I recommend that they rewrite sections of the manuscript to highlight them. At the moment, it comes across as an explanation of a collection of disparate techniques and methods. The authors started by talking about TeamTat and its use in team annotation projects, then launched into explanations of BERT, BioBERT without adequately explaining what their original contributions and the justification for this manuscript is. I recommend tying together all the disjointed pieces and presenting your approach as a cohesive whole.
In addition, some minor edits:
Page 1 Line 11: Do you mean "automated selection of target terms"?
typo in Tuning instead of Tunning
Lines 48 and 49 Page 2: Annotation instead of Annoation
Page 2 Line 63: Link is not working
Page 4 Figure 3: Please provide reference for BioC
Author Response
Please see the attachment. Thanks
Author Response File: 
Author Response.docx
Reviewer 2 Report
This paper is about ["Biomedical Text NER Tagging Tool with Web Interface for 2 Generating BERT-based Fine-Tuning Dataset"]. Unfortunately, the interface is not working. I tried from multiple browsers and asked good friends. It is disappointing that I didn't get to experiment with this tool and test-drive it to see what it is capable of. Perhaps for future submissions, the authors should focus on the novelty of methods as opposed to unavailable service.
Comments for author File: Comments.zip
Author Response
Please see the attachment. Thanks
Author Response File: Author Response.docx
Reviewer 3 Report
Authors have made immense efforts to develop a tagging tool using machine learning and using BERT approach for tagging the named entities and relation extraction. Although authors claim that this approach captures entities that are not identified using NER through BERT architecture, there are some major concerns that needs to be addressed and requires modification.
1. Several Named entity tagging tools and BERT based systems are available, while the authors have chosen to describe three of the related works Textpresso Central, ezTag and TeamTat. In the introduction section, workflow of these systems is explained while authors approach is not explained in detail. A workflow of the developed system would be of much interest to the user rather than the description of the other systems.
2. In the introduction section, please detail the description of the current work.
3. In methods section, BERT based system is described and Input embedding of BERT is shown in Figure 6. As this approach uses a BioBERT embedding, authors could show an illustration of the BioBERT rather than citing BERT embedding from the reference.
4. NER dataset generation is described here and the process of Relation extraction should be described further in more detail.
5. Results section needs improvement and should be described in more detail. Performance evaluation of the proposed method should be done in comparison with existing methods.
6. To further explore the performance of the proposed method, the results shown in the web service implementation can be illustrated as the case studies for particular disease such as Alzhimer’s disease.
7. Authors have proposed a machine learning-based tagging tool that enables 335 automatic tagging in specific domains, user-defined tagging, and dataset generation for 336 fine-tuning. A detailed note on the machine learning process implemented in this technique needs to be described and performance comparison needs to be implemented.
8. Discussion section only demonstrates only the description of the proposed system and its limitations. A more detailed discussion on the results and the system evaluation would be of interest to the users.
Author Response
Please see the attachment. Thanks
Author Response File: Author Response.docx
Round 2
Reviewer 3 Report
Authors have made efforts to make the performance comparison of existing tools and also have described the BioBERT in more detail.
Overall, the content and language has been improved by the authors with several changes, but there are still minor issues with syntactical constructions and formal scientific writing (e.g. in the added/extended text sections such as ("significantly increasing the amount of data curation - line 50" and "The article used in the example is research on Alzheimer's disease - line 313"). I suggest authors to rephrase these terms and similar terms in the manuscript.
Author Response
Please see the attachment. Thanks
Author Response File: Author Response.docx
