Empowering Large Language Models to Leverage Domain-Specific Knowledge in E-Learning

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Proofread for grammatical/mechanical errors (e.g., p. 2: score are widely used metricS); watch your use of tenses (past/present/future). Write out Bert Score consistently

p. 2: Spell out NLP first time of its occurrence

p. 2: BERT-Score paragraph has repetitive information so remove duplicative information; maybe combine with the following paragraph too.

Slim literature review

p. 2: define/explain BLEU and ROUGE

p. 3: define/explain Llama 2

p. 3: avoid you/your' incomplete sentence in first paragraph 3-1.

p. 3: omit space line in 3-2 and 3-3; format box: customer's

Fig. 1: it looks as if the Q could go directly to Q+relevant content without going through embedding, etc. Is that accurate? If so, you need a decision box so the system knows which path to take. 

p. 4: omit 3-5 3: Semantic Overlap-Based Metrics.

p. 4: define F1 score

p. 5: Table 1: what is the range of BERT scores? What does a score near 1 mean? Explain terms Precision/Recall/F1

p 6:  .... are all well  >> are all good; same issue two lines down; ... newly published paperS... known conceptS

Place appendix AFTER references

The main variable in the paper is your providing the information rather than relying on the pre-existing information. It would be useful to do a content analysis to drill down to the differences between pre-existing information and the enhanced one. 

Comments on the Quality of English Language

generally good writing; see specifics in author notes

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This article is generally well written and approaches an extremely interesting and up to date topic.

However, before it can be accepted for publishing, there are some issues worth mentioning, with the aim to bring some general improvements. I will list them in the order of appearance:

In the Introduction section it may be appropriate to further emphasize the arguments a little more on what exactly this paper brings new in comparison to what other researchers did on similar topics. From this perspective, more titles in the References section would be beneficial, as it would serve to better underline the novelty of this approach.

Page 2, Paragraph 2 – the authors wrote: “… is a technique in NLP that allows LLMs like ChatGPT…”. Please state first what NLP stands for, then use the acronym, as some readers may not know its meaning in this context.

Page 4, Fig. 1 – at the name of Figure 1, please see it again, as the last written words are: “Retrieval Augment Generation Augment”

Page 5, Table 1 – please explain what formula or tool did you use to calculate the BERT score (Precision, Recall, F1 Score).

In the Conclusion section, please explain in more detail the practical implications of your approach. Also, as you wrote: “We analyze the strengths and limitations of our approach, discussing potential avenues for future research”, please include these aspects in the paper as they are mentioned in the Conclusions, but not in the body of the article.

I wish the authors best of luck in their future researches.

Comments on the Quality of English Language

The English language looks fine.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

In this paper, the authors present an approach to empowering LLMs with domain-specific knowledge in the E-learning domain. However, the given methodology is not tested, there is no practical evaluation and use. In addition, the results in Section 4 need to be proved how they calculated. A tool needs to be implemented based on your methodology. Then, you should use it in various case studies and evaluate the results. 

 

Comments on the Quality of English Language

In p.3, Section 3, paragraph 2: detail -> detailed 

In p.3, Section 3.2: You should remove the empty line after the first one. 

In p.5, the mentioned research paper needs to be cited.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The paper has been improved, most of the issues have been addressed. Thus, the paper may be accepted for publishing. I wish the authors best of luck in all their future research.

Comments on the Quality of English Language

The English language looks OK.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors addressed the comments of the previous review.