I'm reading a paper about DBSCAN implementation on Map-Reduce. In preprocessing step, it wants to divide dataset and it says:
We consider two strategies in this paper for different scenarios. The ﬁrst one is given the number of computing slots N and its grid decomposition (N = a ∗ b), it could ﬁnd an approximate division in raw spatial data.
I don't get it exactly. Does he want to get coefficients of number of processing unit and use them to divide different dimensions of data?
P.S: Please help me choose appropriate tags.