akheli / epredictor Goto Github PK

Review the code by:

1. adding comments to general steps

2. removing all unnecessary comments and printing lines

to make the code professional and easily understandable.

Create a method that takes and returns a list of Graphs plus the list of Attributes to compute. This method should add the calculated attributes into the returned Graph.

GN & CONGA & COPRA: fonction vide, throws exception
CONCLUDE: ne retourne pas les graphes mais null
SPLA ne détecte rien pour le moment

Implement SGCI evolution identification method instead of Palla's.

Generating random graphs using GraphStream Generators explained in: http://graphstream-project.org/doc/Generators/

Splitting can give same edges in one snapshot, turn this information into weight for edges then use this it to calculate cohesion.

Allow to visualize the distribution of timestamps a given benchmark, to make it possible to choose duration length for each Snapshot, to make the splitting more effective.

Insert community detection methods (besides CPM and Louvain) with a small Main class for each and test them.

Add attributes of context

Include Louain's detection method

Split method should be able to take the length of periods in parameters to allow splitting by periods length instead of the number of equal snapshots.