Semantic Scholar is the Allen Institute’s search engine for scientific literature. I recently stumbled across a useful feature, it uses AI to summarise papers to TLDR format (short for Too Long, Didn’t Read).
This automatic feature generates a 20 word summary of a paper that uses an extreme summarisation method, effectively picking out the most important information from the title, abstract, introduction and conclusion.
The idea is that instead of reading the paper abstract or skimming the conclusion section to determine if you want to read the paper in full you can read a short 20 word summary instead.
The method is available here: https://github.com/allenai/scitldr published as ‘TLDR: Extreme Summarization of Scientific Documents’ at https://doi.org/10.48550/arXiv.2004.15011.
If you want to try it out for yourself head over to https://www.semanticscholar.org search for a paper and look for the TLDR, for example:
Reading a paper nowadays is like reading a condensed textbook, they are often crammed full of terminology, detailed figures and methods. This ‘information overload’ can make it difficult to pick out the underlying research story and rationale.
All of this begs the question, why do we write papers this way in the first place!?
There are now more papers out there than you can read in a lifetime – the number of papers published each year is increasing and the material is becoming more and more complex.
Maybe it’s time to rethink the format of academic papers.
What do you think?