Incorrect behavior in separate_mulitple_summaries function about booksum

[jigsaw2212]

Hi @dojoteef. Thanks for raising this issue. I'm looking into this at the moment. The regexes that we use to splitting the aggregate summaries are not exhaustive and at times we had to manually intervene and modify the lines in the text, rather than having a regex for every outlier scenario. I am adding some more checks to try and split as many book chapters as possible from this list you've shared. Thanks again!

[jigsaw2212]

My latest commit should've fixed this issue. I have tested out the fix for all sources, and should help with splitting some of the book chapters that we were not able to get before.

[dojoteef]

Thanks so much for looking into all these issues I've brought up! I've shifted focus for now, but will be coming back to the BookSum dataset in the near future and will try to see if I encounter any additional blockers.

Thanks again for all the help!