Integrating Frequent Itemsets Mining with Relational Database

Frequent itemsets mining is becoming increasingly important since the size of databases grows even larger. Currently database systems are dominated by relational database. However the performance of SQL based data mining is known to fall behind specialized implementation and expensive mining tools being on sale. In this paper we analyzed a famous frequent itemsets discovery algorithms FP-growth, and propose a new implementation approach called DBFP-Growth to create disk-based FP-tree based on ORACLE PL/SQL, it can execute faster than using SQL directly. Also presents a novel SQL-based method DRRW, which can remove duplicate records from database without temp table generation.

1 Introduction

Extracting valuable rules from a large set of data has attracted lots of attention from both researcher and business community. This is particularly driven by explosion of the information amount stored in databases such as Data Warehouses during recent years. In business world, many organizations begin to apply data mining techniques directly to raw transaction data Aand some results such as unidentified buying patterns and credit card fraud indications are widely recognized.