Abstract: Compared with a priori-like algorithms, algorithms based on vertical data structure and lattice theory presented by Zaki can improve the running speed with an order of magnitude, and these algorithms are very suitable for mining low support and long patterns. The Eclat algorithm is used as the prototype for discussion of the problem of association rule mining with item constraints. It is proved in theory that with the introduction of item constraints, Eclat+ could greatly improve the efficiency and speed of association rule mining, and the Eclat+ is compared with some other related algorithms.