SBASS: Segment based approach for subsequence searches in sequence databases

This paper investigates the subsequence searching problem under time warping in sequence databases. Time warping enables to find sequences with similar changing patterns even when they are of different lengths. Our work is motivated by the observation that subsequence searches slow down quadratically as the total length of data sequences increases. To resolve this problem, we propose the Segment-Based Approach for Subsequence Searching Technique (SBASS), which modifies the similarity measure from time warping to piece-wise time warping and limits the number of possible subsequences to be compared with a query sequence. That is, the SBASS divides a data sequence X^→ and a query sequence q^→ into piece-wise segments and compares q with only those subsequences which consist of n consecutive segments of X ^→. Here, n is the number of segments in q^→. For efficient retrieval of similar subsequences, we extract feature vectors from all data segments exploiting their monotonically changing properties, and build a multi-dimensional index. Using this index, queries are processed with four steps: (1) index filtering, (2) feature filtering, (3) successor filtering, and (4) post-processing. The effectiveness of our approach is verified through experiments on synthetic data sets.