The semantic core was collected in three iterations.
In each iteration, a cleaning was made from waste words.
At the final iteration there was a clustering.
Clustering and cleaning was done manually.
CSs were gathered from Google Planner and other open / closed bases.