Mining multi- attribute associate rules based on attribute union

Mining associate rules is an important research topic in Data Mining. It is a NP-hard problem to estimate whether there are frequent items which has t-attribute and σ - confidence in database. The important research fields of mining frequent items is to reduce the number of scanning database to improve the algorithm efficiency, the attribute Union theory is proposed to calculate the frequent items in database to improve the data mining efficiency, the main idea is translating the scanning database to find the attribute union, dropping the database, obtained the k frequent item by k-1 item operation using the attribute union, not need scanning database, improve associate rules mining algorithm efficiency; an example is presented.