基于“新颖度”的关联挖掘算法An Association R

关联挖掘的目的是从大量数据中发现对用户有用、新颖、重要的关联规则。传统的关联挖掘算法会产生大量对用户而言显而易见的平凡规则，使那些真正对用户有用的新颖规则被淹没，而一些针对新颖性的改进算法往往又存在先验知识表达复杂且工作量极大的问题。在本文中，我们运用简单的分类树，引入“新颖度”的概念，对Apriori 算法进行改进，得到了基于“新颖度”的关联挖掘算法，此算法既充分考虑了挖掘过程中得新颖性问题，又克服了先验知识表达过于复杂的困难。
关键词：关联规则挖掘；Apriori 算法；新颖度

Abstract: The objective of mining association rules is to find useful, novel and important association rules from large database. Traditional association rule mining algorithm may often produce too many obvious and non-novel rules to user, making really novel and interesting rules submerged. Some former way to solve novel problem need a lot of extra work to preprocess the data. In this paper, we introduce the concept of “novelty” and improve the algorithm into a new association algorithm based on “novelty” to overcome the above problems.
Key words: Association Rule mining; Apriori Algorithm; Novelty