Highway Construction Safety Analysis Using Large Language Models

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study delved and demonstrated the application of LLM through GPT 3.5 API in health and safety of highway construction workers. The introduction is well-written. However, there is no literature review to back the analysis. The authors are therefore advised to:

 

1) Produce a literature review on high construction health and safety, LLM, through GPT3.5 etc. This would strengthen and justify the unit of analysis and premise behind the study. 

2) There are three main types of LLMs and the authors can discuss this in the literature review section. This will support the justification of the chosen approach.

3) Section 2 can now become section 3. 

4) There is no real discussion of table 6. Table 6 must lead to discussion the contribution to knowledge and practice. This can be in a separate section. 

5) Authors must consider restructuring the overall layout of the manuscript.

 

 

Comments on the Quality of English Language

The manuscript would benefit from another round of proofreading and formatting. 

Author Response

Please see the response in the pdf file. thank you

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

See annotated paper for minor corrections.

The methodology overview (Section 2.2) needs to be enhanced to show who (or what) is reponsible for the various steps. Example, it is not clear who should answer the prompts (Figures 6, 7, 8). Also, was this done for all the 1032 cases from the database? Who did this? From these comments then, the effort and knowhow of the user who compiles such an analysis are not clear. The authors are encoraged to be clearer about the execution process.

Some comment is needed on the effort required to answer the prompts and to perform the analysis.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Limited comment with small number of comments: See annotated document.

Comment: It is difficult to read a apper where abbreviations are constantly used. The reader is obliged to go back and look up the abbreviation at the first use. This makes for difficult reading.

Author Response

Please see the response on the pdf file. thank you

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The paper uses LLM for construction safety analysis. Overall, the paper is well-written, but the following comments should be addressed:

1. The methodology section should be improved for clarity. It is hard to follow what exactly was used with sufficient details.

2. Line 191: What does the (E) mean in a Transformer Encoder (E)?

3. Line 196: The title of 2.4 should be ‘Clustering Embeddings’ instead of ‘Calculating Embeddings.’

4. Line 199-204: The paragraph needs to be rephrased. Did the authors use the K-means algorithm or not? Rephrase it in a way so that readers know how to reproduce the embedding clustering instead of introducing K-means.

5. In Tables 2 and 4, Cluster 1 is ‘Stuck by Vehicle’. Should it be ‘Struck by vehicle?’

Comments on the Quality of English Language

The English is good overall. 

Author Response

Please see the response on the pdf file. thank you

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have revised the manuscript accordingly and I will recommend an acceptance in current form. 

Comments on the Quality of English Language

The referencing style may be revised for a final time. The English language is acceptable. 

Author Response

Reviewer 1:

 

Comments:

 

• The authors have revised the manuscript accordingly and I will recommend an acceptance in current form.

 

Response: Thank you. We appreciate the time and commitment reviewing our work.

 

• The referencing style may be revised for a final time. The English language is acceptable.

 

Response: We revised the references.