ERAPN, an Algorithm for Extraction Positive and Negative Association Rules in Big Data

摘要

Over the past two decades, the extraction of positive association rules is a significant research area in Big Data. While the negative association rules has been received a lot of attention, they remained in the shadows due to the difficulty of their extraction. In this paper, we propose Erapn, an algorithm for extraction of valid positive and negative association rules. Frequent patterns can be derived in a single pass to the database, because, of the new technique support counting, called reduction-access-database. As for the generation of potential valid association rules, we introduce a new technique, called reduction-space-rules, by dividing the space candidates into two. Only half of the candidates have to be studied through this technique. Some experiments will be leaded into such reference databases to complete our study.