Here is the source code of the Fast-Skyline algorithm described in (van Leeuwen & Ukkonen, SDM 2015). This implementation is intended to demonstrate the algorithm with the seed selection problem (influence maximisation), but with small modifications it should work for other purposes as well.
The implementation is experimental, and may or may not work for serious use. It is available under the MIT license.
ca-HepTh.txt.gzinto some directory to obtain
./make_db.sh SOMEDIR/ca-HepTh.txt 200 0.01 SOMEDIR/ca-HepTh.jsonThis will pre-process the graph so that we can use the efficient influence estimation heuristic (Static Greedy) proposed by Cheng et al (CIKM 2013). This example will use 200 samples and a constant propagation probability of 0.01 on every edge. The output is stored in
./find_skyline.sh SOMEDIR 50 output.txtThis will read *all*
SOMEDIR(therefore, make sure there is only
SOMEDIR), and run Fast-Skyline as well as the regular greedy methods. Output (the skylines) is stored in