Just networkx. I am building my own algos rn, because the algo in that paper requires to store a bunch of random walks in the DB and keep them updated at all times.