Research On Association Rule Mining Based On Master-Slave Relational Data Model

Data Mining is a new intercrossed subject relevant to artificial intelligence and database. As an important pattern in data mining, association rules mining attract many researchers, and have achieved fruitful results. The majority of these results are the study on association rule mining algorithm which to be aimed at single table. However, With the extensive application of relational database, many of the practical data stored in multiple tables in accordance with the principles of standardization, how to mining association rules from these tables which have semantic contact has been the important subject in this area. In this paper we will discuss mining association rules in multi-relation data. Contributions of the dissertation are as follows:1. The concept, requirements of multi-relational association rule mining and relational database model are analyzed deeply. The ideas, principles, advantages and disadvantages of kinds of multi-relational association rule mining algorithms are probed into.2. For the disadvantages of multi-relational association rule mining algorithms, propose a model of master-slave database, which based on the relation between tables, and a virtual connection method——tuple ID retrorse propagation is lead into data tables, formed algorithm TIDRP. This Algorithm improve the time and space performance problem which the current multi-relational association rule algorithms face.3. Based on the algorithms, the prototype system which accord to the model of master-slave database is constructed.