Efficient String Mining under Constraints via the Deferred Frequency Index

Abstract

We propose a general approach for frequency based string mining, which has many applications, e.g. in contrast data mining. Our contribution is a novel algorithm based on a deferred data structure. Despite its simplicity, our approach is up to 4 times faster and uses about half the memory compared to the best-known algorithm of Fischer et al. Applications in various string domains, e.g. natural language, DNA or protein sequences, demonstrate the improvement of our algorithm.