A Demonstration of SearchonTS: An Efficient Pattern Search Framework for Time Series Data.

In recent years, time series data are everywhere across different industry, which creates a huge demand on time series data analysis, such as pattern search. Meanwhile, it is increasingly realized that only when pattern search results together with information from relational tables could be used in a programming-free way, can they perform analysis on time series conveniently. Hence, casual users highly demand that queries involving pattern search could be performed via SQLs. However, existing database products supporting time series data type lack the capability to perform pattern searches on time series data. This paper presents SearchonTS, an extendable framework for in-database pattern search on time series data. It provides a series of interfaces so that time series index and pattern search can be added and performed in a uniformed and query optimized manner. SearchonTS is implemented as an extension on Informix, which is a database product in IBM software product series. It targets a future release of IBM Informix. We have implemented index-based pattern search for Euclid Distance(ED) via SearhonTS to demonstrate its usability for developers. And real scenario is also provided to show SQL involving pattern search so that users can have a more clear experience of the convenience.