Boris Dyakov
@bojo
Gm fellow casters Who here is working on NLP-based annotation and curation of scientific literature to create queryable knowledge graphs? I wanna jam on something
1 reply
0 recast
0 reaction
cdt
@cdt
@hero not sure if related to your arXiv project from the past at all
1 reply
0 recast
0 reaction
TheHeroShep
@hero
Yeah definitely similar to something I’ve worked / currently working on; @bojo what’re you thinking?
2 replies
0 recast
0 reaction
Boris Dyakov
@bojo
Still a pretty unpolished idea - a vision, even. But one that's been with me since I started grad school: I do functional proteomics work (protein-protein interactions in human cells) and the output of my assays is gene lists upon gene lists. A fun, but tedious aspect of my work is trying to understand [con'td]
1 reply
0 recast
0 reaction
Boris Dyakov
@bojo
..how these genes are connected, and what sort of functional significance might exist around this. We use clues from biomedical literature to come up with new functional/mechanistic hypotheses.. and then do (more/other) experiments to explore these. This is fun, but very inefficient, and relies....
1 reply
0 recast
0 reaction
Boris Dyakov
@bojo
on the ability of scientists (& our cell+molecular bio, genetics, and biochemistry intuition and background knowledge) to find these connections. There's SO MUCH literature out there, and a product that lets you discover these links and suggest which gaps need to be filled would be incredibly powerful.
1 reply
0 recast
0 reaction
Boris Dyakov
@bojo
I'm sure there must be ppl working on this, but since I'm very much an experimental biologist I have no idea where to even start. I should really be focusing on analyzing my dataset for my paper and thesis (trying to finish my PhD asap) but with all the recent AI advances I keep coming back to this idea 😅
0 reply
0 recast
0 reaction
