The cells from the second table are attached to the first table. If selected, the node avoids creating duplicates from the second table. (However, if there are more rows in the first table than the second, this is not possible.)
If the user chooses a string column from both data tables, allocation depends on the values of these two columns. If possible, the new cells are chosen so that the columns match. (If there are no matches in the second column, we simply choose a random row.)