Research On Association Rule Algorithm In Data Mining

"Data explosion, poor knowledge"is a serious problem in the information age, Data Mining is a very useful method to solve the problem. Data Mining is defined as finding knowledge from database, which is a kind of process that reveals potention useful knowledge from massive data. It can find the hidden, previously unknown information which has the potential value when making decisions. So researching on Data Mining technology is very significant in the field of application.This dissertation studies mainly on mining algorithm of association rule,discusses the classical mining algorithm,Apriori algorithm,introduces its basic principle,extisted drawbacks and bottlenecks. Meanwhile, several improved algorithms , such as Sampling algorithm, Partition algorithm and Hash algorithm, are introduced for overcome this bottlenecks. Apriori algorithm has two main drawbacks. One is producing most of candidate item sets in the process of mining. The other is that the scale of database remains unchanged in the course of mining. Then based on this drawbacks, three methods have been introduced in order to decrease the times of scanning database.1. The algorithm of reducing candidate item sets. Large number of candidate sets will be produced in the course of mining, new algorithm can reduce candidate sets so that decrease the number of accessing database.2. he algorithm of reducing the scale of database. With deepening of mining, some records in database are not needed any more, so these records can be deleted. As a result,the scope of database will be decreased gradually.3. Vertical data algorithm. While scanning database for one frequent item sets, data format can be converted from standard style to vertical style.