Transform-based similarity methods for sequence mining
Tara A. Halwes · Digital WPI · 2000
We investigated transform-based algorithms for performing similarity queries in sequential databases. The transform-indexing method of Agrawal, Faloutsos, and Swami (1993) was evaluated, using both Fourier and Walsh transforms to extract features. We extended the method to enable it to handle categorical data as well as different notions of distance. Analytical and experimental evaluation of these methods was performed, using financial data and simulated genetic data.