Interactive and space-efficient multi-dimensional time series subsequence matching

Abstract

We develop a highly efficient access method, called Delta-Top-Index, to answer top-k subsequence matching queries over a multi-dimensional time series data set. Compared to a naïve implementation, our index has a storage cost that is up to two orders of magnitude smaller, while providing answers within microseconds. Additionally, we apply cache optimization techniques to speed up the construction of the index. Finally, we demonstrate the efficiency and effectiveness of our technique in an experimental evaluation with real-world data.