Selections = data driven restrictions
As we have already mentioned previously, the only operation on data similar to the relational model is selection. Selection is normally based on some predicate involving data values. The predicate would usually specify some restrictions on the set of cells whose values are to be examined. As in the relational case, this restriction logically has nothing to do with the resulting data set.
Sample query: look at the slice of Q via m Î D1 and if the measure x has positive value, select the fifth coordinate of the corresponding cell. The selection in this example is thus made within D5. Note that although we are examining cells, we are selecting slices. We would have to use two queries in order to achieve the same in the relational case.