/DenseTrussSearchKT

the code implemented for a paper "Cohesive Subgraph Search Using Keywords in Large Networks".

Primary LanguageC++

DenseTrussSearchKT

This is the code implemented for a paper Cohesive Subgraph Search Using Keywords in Large Networks. Originally written by others but modified by me.

Compilation Note

This program needs SNAP library to be installed on your computer as a LIBRARY. The compiler tries to find libSnap (compiler flag: -lSnap) in your computer. If not, make will exit. Also, this program assumes that Snap library is copied to the following paths:

  • /usr/local/include/snap-core

  • /usr/local/include/glib-core

  • /usr/local/include/snap-adv

  • /usr/local/include/snap-exp

If this is NOT where SNAP Library is located on your computer, you are free to modify the Makefile.

Possible Improvements

Several parameters in this piece of code are chosen RANDOMLY. You are invited to check what parameters work the best:

  • maxOutputHeapSize in BanksAlgorithm.hpp, which is the minimum size for BANKS algorithm to stop searching for new trees. Bigger size means better results (maybe?) but lower speed.

  • Deleting policy of early-stop batch based deletion in MinDenseTruss.cpp. Currently, my implementation starts the deletion size with 1 and increases it by 1 when the last deletion was successful. Otherwise it is reset to 1. There may be better ways to handle this.

  • earlyStopThreshold in MinDenseTruss.hpp, which is the number of unsuccessful deletion it requires for the algorithm to stop deleting.

  • steinerTreeThreshold in MinDenseTruss.hpp, which is the threshold of the size of vertices added to the steiner tree on once in the local exploration algorithm.

If any bugs exist, report to me or you can fix them by yourself. There are extensive comments so it should be fairly easy. Enjoy!