codeseg 1.0.3

CoDeSEG Community Detection Package

Community Detection in Large-Scale Complex Networks via Structural Entropy Game (CoDeSEG)

CoDeSEG supports undirected, directed, weighted, unweighted, overlapping, non-overlapping, and dynamic community detection. The relevant parameter descriptions are as follows:

Parameter	Description	Type	Default	Required
in_path	Input file of graph edge list	file	None	Yes
out_path	Output file of communities	file	None	Yes
ground_truth	Ground truth file	file	No	No
weighted	Weighted graph	bool	false	No
directed	directed graph	bool	false	No
dynamic	dynamic graph	bool	false	No
overlap	Overlapping communities	bool	false	No
gamma	Overlapping detecting factor	float	1.0	No
tau	Non-overlapping entropy threshold	float	0.3	No
r	Stable round threshold for dynamic detection	int	2	No
it	Maximum number of iterations	int	10	No
parallel	Number of threads	int	1	No
verbose	Print detection iteration messages	bool	false	No

Note

Build Requirements

• CMake >= 3.22
• python >= 3.8

data format

The format of the input edge list is as follows:

   1 \t 2 \n
   1 \t 3 \n
   2 \t 3 \n

For dynamic graphs, the input should be a file containing edge lists of multiple network snapshots, stored in the /data/ntwk directory. The file structure is as follows:

data(your dataset)/ 
├── ntwk 
    ├── 1.txt
    ├── 2.txt
    ├── 3.txt
    ...