Study Of Incremental Updating Technology For Negative Association Rules

Recent years, with the popularity of computer, the Internet and the development of database, the databases in various application fields have accumulated lots of date. Through data mining analysis and understanding of these data, which reveals the hidden useful information, and become the most active areas of research. Mining Association Rules of data mining is an important model, and has important theoretical value and prospects for a wide range of applications.Association rules on the data can have positive and negative association rules on the correlation.Recetly, mining positive association rules has a widespread concern. It did not give sufficient attention for contain negative or negative attributes of the project association rules. However, in many application areas, the negative things are also very important factors of the sources of information, therefore, it is necessary to study things between the negative and associated attributes relations.On the other hand, with the time flying, data in the database will change. This is what we call incremental updating problems. General sense of the incremental updating problem can be understood as: to increase or reduce data in the original database, and then to update the rules of association in the new database.At present, the research on the incremental updating for association rules is mainly for the negative association rules. For example, Agrawal R, and Srikant R proposed FUP algorithm; Brin S, Motwani R, and Silverstein C proposed FUP2 algorithm; Feng Yucai and Feng Jianglin proposed IUA and PIUA algorithms in the homeland. The research on the incremental updating for negative association rules is relatively small. And the generalized incremental updating is divided into: changes in the database and the changes of min- support and the min- confidence.There are some differences between the increcmental updating for the positive and negative association rules. Specific performances are the following:①Positive association rules only exist in the frequent itemstes, but the negative association rules not only exist in the frequent itemsets, but more exist in the infrequent itemsets;②Positive association rules only have one form(A=>B),But the negative association rules have three forms(┓A=>B,A=>┓B,┓A=>┓B);③In resolving the positive association rules, it is necessary to update the frequent itemsets in the new database, and the positive association rules can be calculated by the formula; When resolving the issue of incremental updateing for the negative association rules, it is necessary to find all of the frequent and infrequent itemsets, and Mining algorithms will have to use to mine the negative association rules.It is mainly introduced in the thesis that Data Mining Theonology, association rules, the research on the classic algorithms of association rules and incremental updating.The research work of this thesis offers certain theoretical foundation and methods effectively for further research to the existing association rules and incremental updating for association rules.