Distributed Mining of Partial Periodic Patterns in Sequences

Abstract

It has been an important task of discovering frequent fragments as particular patterns from large sequence databases generated from a variety of applications. In general, the patterns to be discovered may partially and asynchronously exist in sequences, and even contain gaps. In addition, it is necessary to collect the information regarding the locations and frequencies of the patterns. How to enumerate candidate patterns for evaluation without exponentially increasing the computation is another problem. In this paper, the modified periodicity transform is proposed to meet the requirements mentioned above. Also, a distributed computing framework is implemented to perform the mining task more efficiently. Both synthetic and biological sequences are utilized to examine the approach. The experimental results demonstrate the efficiency and effectiveness the system.