Matt Stabeler

We took a limited period in the Enron dataset, and split it into two parts, a training phase, and a testing phase:

Activity in the Enron dataset, hourly, daily, weekly, showing possible periods that could be used for training/testing

In this case, the first period  shown in the above plot was used.

For HGCE the comminity structure is shown below:

Community Structure created by HGCE on Enron-split-a

Results of Routing are not good for HGCE:

Results for Enron-split-a dataset

Variation in parameters (threshold) for delivery ratio, on HGCE are shown below:

Benchmark results are shown below:

In which Unlimited flood makes nearly 50% delivery ratio.