Do you have any plans to add a graphical visualization of top/central papers?
https://metacademy.org/graphs/concepts/bayesian_logistic_reg...
and src code for the graph view is here: https://github.com/metacademy/metacademy-application/blob/ma...
they do some clever hiding of edges so graph is not overwhelming, but still only O(100) nodes.... for O(100k) nodes you'll need to do some selection for sure ;)
I love this idea btw, I’m going to use it to find some holiday reading!
But long story short, I end up doing most of the graph analysis by passing in the citations, using PyO3, to graph-tool in python then returning the data I need about each paper. I am planning on moving that over to Rust. But not being an academic I wanted to get feedback on the quality of the results before making it difficult to quickly test different types of algorithms.
The biggest hurdle was the speed of the graph creation. Basically taking a 250,000,000 paper/2,500,000 citaiton db and creating graphs that could be up to 200k papers and 3-4mm citations. For that I ended up learning/using Rust (which was a great experience).
The plan is to keep it totally free and hopefully get some institutional support once I get a better handle on demand and costs.
Ask me anything!
EDIT: As you are going through the site, be sure to use the purple "+" buttons to create your own graphs centered on the topic of your choice. That combined with the in-graph keyword filters are probably the most powerful ways to quickly zero in on the most relevant literature.