A Parallel Algorithm for Mining Density-Aware Distinguishing Sequential Patterns with Spark

摘要

Distinguishing sequential pattern (DSP) mining is a useful technique to discriminate a set of sequences of one class against a set of sequences of another class. One kind of DSP that, considers the density concept in DSP mining (called density-aware DSP) has many applications in bioinformatics and computational biology. However, the previous method to mine density-aware DSPs suffers from the inefficient density computing. As a result, the previous method cannot deal with the datasets with large scale. To break this limitation, we design and implement a parallel mining method to discover density-aware DSPs using Spark, which is a popular framework for parallel computing. Our empirical study on real datasets demonstrates that our proposed method is efficient and scalable.