Users can control the characteristics of generated graphs by acMark.
- numpy >= 1.14.5
- scipy >= 1.1.0
$ python acmark.py
'test.mat' is generated by the example.
In a python code,
import acmark
acmark.acmark(output='test.mat',n=1000,m=4000,d=100)
Other parameters are described below:
outpath : path to output file (.mat)
n (=1000) : number of nodes
m (=4000) : number of edges
d (=100) : number of attributes
k (=5) : number of clusters
k2 (=10) : number of clusters for attributes
alpha (=0.2) : parameters for balancing inter-edges and intra-edges
beta (=10) : parameters of separability for attribute cluster proportions
gamma (=1) : parameters of separability for cluster transfer proportions
node_d (=0) : choice for node degree (0:power law, 1:uniform, 2:normal)
com_s (=0) : choice for community size (0:power law, 1:uniform, 2:normal)
phi_d (=3) : parameters of exponent for power law distribution for node degree
phi_c (=2) : parameters of exponent for power law distribution for community size
delta_d (=3) : parameters for uniform distribution for node degree
delta_c (=2) : parameters for uniform distribution for community size
sigma_d (=0.1) : parameters for normal distribution for node degree
sigma_c (=0.1) : parameters for normal distribution for community size
r (=10) : number of iterations for edge construction
att_ber (=0.0) : ratio of attributes which takes discrete value
att_pow (=0.0) : ratio of attributes which follow power law distributions
att_uni (=0.0) : ratio of attributes which follow uniform distributions
att_nor (=0.5) : ratio of attributes which follow normal distributions
dev_power_max (=3) : upper bound of deviations for power law distribution for random attributes
dev_power_min (=2) : lower bound of deviations for power law distribution for random attributes
dev_normal_max (=0.3) : upper bound of deviations for normal distribution for random attributes
dev_normal_min (=0.1) : lower bound of deviations for normal distribution for random attributes
uni_att (=0.2) : range of paramters for uniform distribution for random attributes