News for hackers. Stuff that matters.

A comment on TLDR: Extreme Summarization of Scientific Documents.

Re:TLDR: Extreme Summarization of Scientific Documents by justarandomq on Friday November 27, 2020 @07:36PM (#25233968)
What you get from applying TLDR to their paper:

We introduce SCITLDR, a new multi-target data set of 5.4KTLDRs over 3.2Kpapers.

Keeping pdf's copy-paste artifacts:

We introduceTLDRgeneration, a new formof extreme extreme summarization, for scientific pa-pers.

Adding intro and conclusion (optional):

We introduce SCITLDR, a new data set of 5.4KTLDRs over 3.2Kpapers.

[0] https://scitldr.apps.allenai.org/

[ Parent | on HN ]
Re:TLDR: Extreme Summarization of Scientific Documents by austinjp on Saturday November 28, 2020 @05:58AM (#25236881)
Yeah. I happen to have been looking at this problem in my spare time recently. I tried a bunch of abstractive AIs and approaches, and none produce consistently usable results.

I'm sticking with extractive approaches plus a bunch of hard-coded general and domain-specific rules for now.

[ Parent | on HN ]
Re:TLDR: Extreme Summarization of Scientific Documents by m1sta_ on Saturday November 28, 2020 @05:03AM (#25236643)
Yikes
[ Parent | on HN ]