Data segmentation | Snapshots creation

1 Segmentation of data:

• Social networks datasets/benchmarks can be of different structure and format, EPredictor allows to load any file type formats can be constumised by the user.
• The segmentation of dynamic networks into datasets can be done using:
  • The number of snapshots: the number of snapshots to obtain by the segmentation.
  • Snapshot duration: the duration of a snapshots, it can be given by hours, days, weeks, months or years... format .. All snapshots obtained will be of this duration.
  • List of snapshots: a list of durations can be given. In this case, every duration represents the convering of one snapshot.
• Snapshots can be exported in separated files or collected in the same file.
• Overlapping: the segmentation by default generates distinct snapshots. EPredictor allows though to do segmentation using a value of overlapping between consucitive snapshots. This value represents the percentage of 'unity' between consucutive snasphots. This value is very used by researchs in the prediction of community evolution is social networks because it allows to reduce evolution speed and thus, to better identify community evolution.