FASTEN

FASTEN is a generative model to solve diffusion network inference problem.

Our model infers multiple diffusion networks using the first-hand sharing pattern.

We also provide several diffusion network models listed below.

• InfoPath
• MMRate
• MixCascades

Please refer to my paper for more details.

Uncovering Multiple Diffusion Networks Using the First-Hand Sharing Pattern
Pei-Lun Liao, Chung-Kuang Chou, and Ming-Syan Chen
Proceedings of the 2016 SIAM International Conference on Data Mining. 2016, 63-71 

Dependency

• Openmp
• SNAP 2.3 Library @ http://snap.stanford.edu/snap/releases.html
• Gnuplot

Note:
You should compile the SNAP core, cascdynetinf.cpp and kronecker.cpp before doing installation.

• SNAP core is in <Your SNAP Library Path>/snap-core directory.

  To compile SNAP core.

  cd <Your SNAP Library Path>/snap-core
  make
  

• cascdynetinf.cpp and kronecker.cpp are in <Your SNAP Library Path>/snap-adv directory.

  To compile cascdynetinf.cpp and kronecker.cpp

  cd <Your SNAP Library Path>/snap-adv
  g++ -c cascdynetinf.cpp -o cascdynetinf.o -I ../snap-core -I ../glib-core ../snap-core/Snap.o
  g++ -c kronecker.cpp -o kronecker.o -I ../snap-core -I ../glib-core ../snap-core/Snap.o
  

After compilation put the compiled objective files i.e. Snap.o, cascdynetinf.o and kronecker.o into lib directory.

Installation

Get codes from Github.

git clone https://github.com/plliao/FASTEN.git

cd FASTEN

Set the SNAP library path in Makefile

vim Makefile

Find SnapDirPath = ../Snap-2.3 in Makefile and set your SNAP Library path.

Create the required directories.

mkdir obj bin

Compile codes using make command.

make -j 4

The compiled programs are in the bin directory.

Usage

Please check all model parameters listed in the corresponging cpp file.

We provide an example in exp.sh script for you to refer.

To run the script

./exp.sh

Note that you should create the required directories before you run the script i.e. mkdir plot result data.

Program Descriptions

Models

• InfoPath.cpp: main file of InfoPath model
• MMRate.cpp: main file of MMRate model
• MixCascades.cpp: main file of MixCascades model
• FASTEN.cpp: main file of FASTEN model

Evaluations

• EvaluationAUC.cpp: PRC AUC evaluation file
• EvaluationMSE.cpp: MSE evaluation file
• EvaluationMultiple.cpp: multiple network evaluation file

Utility

• generate_FASTEN_nets.cpp: cascades and network generator using our diffusion model
• DataMerger.cpp: a program to merge several cascades and network files into single file.

Reference

1. InfoPath, Structure and Dynamics of Information Pathways in On-line Media, at WSDM 2013
2. MMRate, MMRate: inferring multi-aspect diffusion networks with multi-pattern cascades, at KDD 2014
3. MixCascades, Cluster cascades: Infer multiple underlying networks using diffusion data, at ASONAM 2014
4. FASTEN, Uncovering Multiple Diffusion Networks Using the First-Hand Sharing Pattern, at SDM 2016