Meta-Apriori: A New Algorithm for Frequent Pattern Detection

Frequent patterns mining is one of the most important data mining techniques, used for extracting information and knowledge from ordinary data. In this paper, we are interested by improving Apriori, which is one of the most used algorithm for extracting frequent patterns. First, we propose some enhancements for the Apriori algorithm. Then we develop Meta-Apriori a new recursive algorithm based on Apriori. As we know, the major drawback of Apriori is its temporal complexity, which makes it difficult to practice. The aim of our algorithm is to reduce substantially the runtime so as, to increase its efficiency while preserving its effectiveness. The main idea is to use the "divide and conquer" technique, which consists in partitioning the whole database into small ones and then applying Meta-Apriori if the database is huge or Apriori if it is of reasonable size. By merging the achieved results, we obtain the outcomes for the whole database.