Are there any discussions comparing the generated documents (D) with the original seed documents?

[txsing]

Hi,

I’ve just gone through your work, and the RAGEval freamwork is really interesting and highly valuable from a practical standpoint. Great job on putting together such a solid piece of work!

I have a few questions I’d love to ask:
1. Are the generated documents (D) considered a rewritten version of the seed documents?
2. If so, what are the most significant benefits of using such rewritten documents for follow-up QA or reference generation?
3. Additionally, how do you control potential issues like information loss or misinterpretation during the rewriting process?

Thank you for your time, and I really appreciate to learn more from your work.