A Study On The Theory And Application Of Dynamic Association Rules

Data mining is an important method of processing data, and association rules mining is one of its key research areas. Many algorithms have been developed for mining association rules, but almost all of them assume that the rules are valid all the time and don't change. These rules are considered as static ones. In fact, it is possible that a rule and the character of data will change greatly with time.Based on the system of general association rules, an improved approach for describing and assessing association rules is introduced in this study. A rule described by this new method is called a dynamic association rule. And a dynamic rule is different in that it contains not only a support value and a confidence value but also a support vector and a confidence vector.The main works of this study are as follows:In Chapter 1, the history of and the main methods in data mining are introduced as well as its applications and development tendency.In Chapter 2, the basis of association rules mining technique is introduced, and three classic mining algorithms are presented.In Chapter 3, based on the system of general association rules, a new conception, dynamic association rule, is introduced, which improves the description and assessment of a rule. Also, two algorithms for mining dynamic association rules are presented as well as several ways for using them.In Chapter 4, a system for study assistance using dynamic association rules and Web text mining technique is discussed, which can gain useful knowledge from network databases to assist research work.In Chapter 5, dynamic association rules mining is used to analyze the data from the fourth distillation plant of Sinopec Maoming Petrochemical Corporation. And the procedures and results of data analysis are presented. The rules found in this process could be used in real production.Chapter 6 concludes with a summary and discussions of future and prospective research on open problems.