CPPG: Efficient mining of coverage patterns using projected pattern growth technique

Abstract

The knowledge of coverage patterns extracted from the transactional data sets is useful in efficient placement of banner advertisements. The existing algorithm to extract coverage patterns is an apriori-like approach. In this paper, we propose an improved coverage pattern mining method by exploiting the notion of "non-overlap pattern projection". The proposed approach improves the performance by efficiently pruning the search space and extracting the complete set of coverage patterns. The performance results show that the proposed approach significantly improves the performance over the existing approach. © Springer-Verlag 2013.