Research Of Association Rules Mining Algorithm In Relational Databases

With the rapid development of database technology, worldwide sharp rise in the volume of data storage, in the face of this challenge, data mining technology came into being, association rule mining, especially association rules mining relational database as one of an important research content, has a wide range of applications.Today, the common methods for data mining in relational databases are through a algorithm to pruduce a number of possible models, and then calculate the supportance of these model in database. if the supportance of model exceed the specified standard,it is the desired model, more aften in this area is Apriori algorithm and some of it's commonly used variant.these algorithm encounter a number of issues when they mining association rules in relational databases: first,along with increased complexity of questions,these algorithm will take unbearable burden to system. secondly,because these algorithm does not fully consider the characters of the association rules mining in relational databases, so it is unable to carry out the result when it face with the relational databases in complex pattern.This article is written in such circumstances, the paper first introduced the process of data mining and it's classification, and then focuses on the relevance theory of association rules and the application in relational database.on this basis, through a combination of inverted index and concept grid technology, give a new relational databases association rules mining algorithm.on the one hand ,this algorithm maintain the simple and easy-to-use character of original classic algorithm Apriori algorithm. on the other hand,through the introduction of the inverted index structure,the number of full-table sacn reduced to once, and then improve the efficiency of the algorithm;through the proposed concepts such as connectivity and some new methods, the algorithm have the ability to mining association rules in more complex relational database; by introducing the concept of concept lattice, the number of database scans in measuring the supportance of model will be reduced, this method further improve the efficiency of the aogorithm. Through the demonstration found that this algorithm has the characters of generitic, fast and efficient.Finally, a summary of the full text of the proposed .This algorithm still has some problems and there are still have many works need to be done in future.