Temporal Association Rule Using Without Candidate Generation

Abstract

Associationship is an important component of data mining. In real world, the knowledge used for mining rule is almost time varying. The items have the dynamic characteristic in terms of transaction, which have seasonal selling rate and it holds time-based associationship with another item. In database, some items which are infrequent in whole dataset may be frequent in a particular time period. If these items are ignored then associationship result will no longer be accurate. To restrict the time based associationship, calendar based pattern can be used [5]. Calendar units such as months and days, clock units, such as hours and seconds & specialized units , such as business days and academic years, play a major role in a wide range of information system applications.[11] Our focus is to find effective time sensitive algorithm for mining associationship by extending frequent pattern tree approach [3]. This algorithm reduces the time complexity of existing technique[5]. It also uses the advantages of divide & conquer method to decompose the mining task into a smaller tasks for database.