/K-EIS

Primary LanguageC++

K-EIS

Methods

In our experiments, all the code implements with C++ code.

Our Methods

Link: SourceCode/OurMethods

  • 1-EIS: K-EIS for single target.
  • K-EIS: K-EIS for multiple target.
  • Mix-EIS: Improved K-EIS for multiple target based on subgraph partition.
  • Mix-EIS-IGS: Mix-EIS includes a single-target searching algorithm, using IGS to replace 1-EIS for comparison.

Compare Methods

Link: SourceCode/CompareMethods

  • IGS: A single-target search method over DAGs.
  • TS-IGS: Improved IGS.
  • BinG: A single-target search method over Tree.
  • KBM-IGS1: A single-target search method over Tree.
  • KBM-IGS2: A multiple-targets search method over Tree.

Datasets&EXP

Five Datasets

Link: Datasets&EXP

  • ProductClassification: Tree.

  • Amazon: Tree.

  • ACM_CCS: DAG.

  • Wiki_Edits: DAG.

  • ImageNet: DAG.

Format: The first line with two integer $n$, $m$ , represent the number of vertices and directed edges in the graph.

The next $m$ lines, each line contains two integer $u_i,v_i$, represent the $i$-th edge from $u_i$ to $v_i$ .

EXP

Put the source code into same folder with datasets, testcase and run.sh.

./run.sh

Single-target :

link: Datasets&EXP/SingleTargetTestCase

Multiple-targets :

link: Datasets&EXP/MultipleTargetsTestCase

For each folder K = x , x represents the number of the hidden targets.