How to conduct t-test in summarization task?

[nlpmc]

Hi~

Thanks for releasing the code of your interesting paper. I found that, in your paper, the t-test has been conducted. I have a question that how to conduct the t-test in the summarization task? Is it correct that calculating the ROUGE score on each generated summary of two methods, and then using the standard t-test script to calculate the p-value?

As you may know, the official ROUGE script is very slow when calculating the score. Do you use the standard ROUGE script to calculate the score on each generated summary and then calculate the p-value?

Thank you so much!