Terminology

 Given a base table, choose three columns,

     * pivot_row
     * pivot_column
     * pivot_value

 such that for each pair of values from (pivot_row, pivot_column),
 there is at most one distinct pivot_value.
 
 A pivot table has :

    * one column whose values are the distinct values in pivot_row
    * other columns corresponding to distinct values of pivot_column
    * the values in those columns correspond to the unique values in 
      pivot_value for the corresponding column and row value.