In order to gain more etticiency when calculating a Datalog program bottom up，a suggestion is proposed to divide the Set of Datalog rules into the several groups．A reasonable sequence is provided for those groups，and SO for rules in fhe same group as well．Consequently．according to those sequences scheduling rules will result in more effi-ciency when calculating a Datalog program bottom-up．This ordering technique can be ap-plied to any traditional bottom-up algorithms，hence the wide application of group．The authors put forward some improved versions of certain traditional algorithms．and they are more efficient than the previous versions in space and time．It is also pointed out that the conception of group has more applications other than improving bottom-up calculations．