Given a graph where each node is also associated with content vector, find communities such that each community is coherent in terms of both structure and content.
The implementation for the paper Laura M Smith, Linhong Zhu, Kristina Lerman, and Allon G Percus. Partitioning Networks with Node Attributes by Compressing Information Flow. In arXiv preprint arXiv:1405.4332.
##Installation 1.Requirement:
g++, gcc compiler
2.platform support:
Mac, Windows, Linux
3.Compiling:
go into either the Top down or Bottom up, type make to automatically compile and generate the executable files
##Usage
+ Bottom up approach
Usage: graphfilename [option]
"Use '-o' to indicate step-by-step output";
"Use '-of' to indicate the output folder"
"Use '-suffix' to indicate an output file suffix";
"Use '-c' to indicate content column";
"Use '-d' to indicate dictionary column";
"Use '-dir' to indicate a directed graph (default is undirected)";
"Use '-tau ' to indicate the teleportation probability (only used for directed graphs, default is 0.15)";
"Use '-g' to indicate there is a ground truth file (Format: NodeNumber \t Community Number)";
"Use '-df' to indicate the dictionary filename";
+ Top down approach
usage: graphfilename [option]
"Use '-o' to indicate step-by-step output";
"Use '-of' to indicate the output folder"
"Use '-suffix' to indicate an output file suffix";
"Use '-c' to indicate content column";
"Use '-d' to indicate dictionary column";
"Use '-dir' to indicate a directed graph (default is undirected)";
"Use '-tau <tau>' to indicate the teleportation probability (only used for directed graphs, default is 0.15)";
"Use '-g' to indicate there is a ground truth file (Format: NodeNumber \t Community Number)";
"Use '-f' to indicate the dictionary filename";
"user '-e' to indicate extended content column";
"user '-trials <trials>' to return the best partitioning of <trials> trials";
##Input format + format for graph
The first line is number of nodes, and starting from the second lins is the adjacence list of each node formated as follows:
node_id,degree_d:neighboreid1,weight1:neighborid2,weight2:...neighboridd,weightd
Note that the node_id is within the range [0,n-1], where n is number of nodes, and the list of neighbors are sorted in ascending order too.
An example of input graph format is as follows:
3
0,2:1,1.0:2,1.0
1,2:0,1.0:2,1.0
2,2:0,1.0:1,1.0
where this graph is a triangle with three vertices
+ format for features
Each line is the feature vector representation for a node formatted as follows:
[node_id] tab [number_of_features d] tab [feature_index1]tab[feature_weight1]...[feature_indexd] tab [feature_weightd]
An example of input feature format is as follows:
0 2 0 5.000000e-01 1 5.000000e-01
1 2 0 5.000000e-01 1 5.000000e-01
2 2 0 5.000000e-01 1 5.000000e-01
3 2 2 5.000000e-01 3 5.000000e-01
4 2 2 5.000000e-01 3 5.000000e-01
5 2 2 5.000000e-01 3 5.000000e-01
6 1 3 1
7 1 3 1
8 1 3 1
9 1 3 1
10 1 3 1
##Output format
Each line i gives the partitioning id for the i-th vertex
##Common Questions