Quasilinear-Time Search and Comparison for Sequential Data

Abstract: This paper proposes a new algorithm for the evaluation of similarity between two sequences in quasilinear time. It describes the theoretical, practical and implementational aspects of the algorithm. The proposed method is a new approach dedicated to the computation of sequential similarity in contrast to other methods like the Jaccard Index which although designed for the computation of similarity of sets have been frequently used on sequences. The method is generalizable and applicable to any form of sequential data of a finite alphabet (binary files, DNA sequences, natural language etc.)