Study On Mining Association Rules Of Temporal Data

The technology of data mining began to spring up in the early 1990s, and quickly grew up to be the focus of research in the field of computer theory and application. Now data mining is a multidisciplinary research area. By using the successful production of database technology, artificial intelligence, machine learning, statistics, information retrieval, high-performance computing and visualization, the data mining technology has succeeded in breaking the situation of "rich data but poor knowledge".With the development of data mining technology, people begin to realize the necessity and significance of temporal data mining. Time is an objective attribute of the objective world and it is the inherent attribute of the data. It is inappropriate that we overlook the temporal semantics of data in the data mining process. So people pay more and more attention to mining temporal knowledge such as temporal models, trends, constraints, causal relationships and so on. Especially, the computer academia pays the most attention to mining temporal association rules.This paper presents a method for mining the periodic temporal association rules of discrete attributes from the temporal database. There are many studies on the periodic temporal association rules such as mining periodic temporal association rules of continuous attributes, mining temporal association rules in fixed cycles and so on. On the basis of the results of these studies, this paper proposes a method for computing the cycle of the discrete attributes from the temporal database, and then mining the periodic temporal association rules by implementing the adaptive Apriori algorithm.The main works and achievements of this article are:1. On the basis of dividing and marking the valid time intervals of the temporal database, this paper presents a method for computing the cycle of the discrete attributes. The detailed steps of the method are described in the paper.2. This paper proposes a new method for marking the time intervals. We use a sign set to replace the time interval by dividing and marking the valid time attribute of the temporal database. That is how to realize the discretization of the valid time intervals. 3. The adaptive Apriori algorithm is presented in this paper. When computing the frequent itemsets, the suited itemsets are regarded as different itemsets if they bring different sign sets. So the support degrees of such itemsets are calculated respectively. Then we can get the frequent temporal itemsets.4. After we get all the temporal association rules, on the basis of the cycle of the discrete attribute that has been calculated, we give reasonable explanation of the rules.5. Analyses the main reasons for the improvement of the performance of the adaptive Apriori algorithm presented in our paper.In this paper, we calculate the cycle of the discrete attribute and realize the discretization of the valid time intervals by dividing and marking the valid time attribute of the temporal database two times. And then we can implement the Apriori algorithm. Something is worthy to be presented is that the adaptive Apriori algorithm proposed in this paper is more efficient by significantly improving the speed of the iteration constringency and reducing the times of scanning the database when computing frequent itemsets.