Searching time series using textual approximation

There has been much research work on similarity search in time series for many years. In this paper, we propose a new time series similarity search technique called TAX, which is using textual approximation method to apply existing document retrieval methods to time series database. The proposed textual approximation is a method that extracts set of terms called T-terms from a time series to approximate time series data using document retrieval methods from a time series database. The paper describes this novel similarity search technique using the textual approximation method, including T-term extraction and use of document retrieval methods. We will show that TAX is effective for classification as well as search in time series data set in our evaluations.