Thinking about the hyperlink thing more, I think I would revise my suggestion to avoid undefined acronyms - I think you could explicitly encourage them, but only if they are tagged and defined and linked to the originating paper so that experts can skim and newcomers to the field can get context with an expandable html footnote or something.
That way you don’t have a zillion tldrs trying to explain what NeRFs are and do, they all just tag NeRF and link to the canonical tldr for NeRF
That would make sense. I think how to approach it, so we can have collectively built definitions of concept like NeRF. Maybe we can treat them in a similar way to TL;DRs - everyone can write their own version and we can vote on the best one?
Another approach could be some collective improving the definition, similarly to how Wikipedia contributors work
Yeah that sounds awesome to me. It addresses one of the shortcomings of the conventional abstract style which is that nothing is cross referenced - most journals still disallow citations in the abstract
A feature like this would be great for discovering new topics. Right now my approach is to find an interesting paper, and if I need to dig into underlying concepts to really evaluate whether to read in detail then I skip to the lit review and hope for some good bread crumbs, or maybe jump to a fresh GScholar query to do some depth-first-ish graph traversal
It would be really nice to just surf some linked tldrs with high level descriptions of papers and concepts instead!
