Hi Wei,
As indicates in line 81 in run_summarization.py: "if len(annot['text']) > 1:". The system summaries with one sentence are ignored when computing correlation. Totally 246 system summaries are ignored. I am curious about the following questions:
1. Why do we need to skip them? I think it is more reasonable to include them.
2. Do you actually want to use ">=" instead of ">"?
3. Was this operation applied to the baselines?
Hi Wei, As indicates in line 81 in run_summarization.py: "if len(annot['text']) > 1:". The system summaries with one sentence are ignored when computing correlation. Totally 246 system summaries are ignored. I am curious about the following questions:
Thank you!
Wang