Similarity Search in High-dimensional Spaces with Applications to Time Series Data Mining and Information Retrieval

Abstract : We present one of the main problems in information retrieval and data mining, which is the similarity search problem. We address this problem mainly from a metric perspective. We focus on time series data, but our general objective is to develop methods and algorithms that can be extended to other data types. We investigate new methods to handle the similarity search problem in high-dimensional spaces. The novel methods and algorithms we introduce are tested extensively and they show superiority over other methods and algorithms in the literature.