Latent Semantic Analysis for Mining Rules in Big Data Environment

Abstract

This Finding valuable rules from a given data set and detecting events using the rules are recent popular research topics. The association rule reduction technique finds unnecessary associations rules and removes them for extracting meaningful relationship between data. The researches on enhancing the reduction rate of final association rules and efficient data structure minimizing the number of scans have been actively performed to reduce the execution time. The previous schemes sometimes fail to reduce the association rules while more reduction is possible since they do not consider the relationship between the data items. In this paper we propose Latent Semantic Analysis (LSA) reduction technique for mining valuable rules at high speed regardless of the number of items. The proposed scheme extracts the relationship such as inverse and equivalence between a set of items. Computer simulation reveals that it significantly increases credibility, support, processing time, reduction rate of the rules and rejection rate of the item, compared to the existing schemes.