The goal of our research is to develop a high performance data mining system (HPDM), which can manipulate very large scientific databases. The research pursues an application-oriented approach with special focus on bioinformatics (e.g., protein structure prediction). The HPDM system is based on a three-tiered architecture consisting of a front-end interface, visualization, and query tool, a middle layer Data Mining Template Library of common high-level mining algorithms and a core set of data mining "primitive operations'', and a back-end Extensible Data Mining System tightly integrated with a database system, and delivering high performance.

• Develop a Data Mining Template Library (analogous to Standard Templatre Library in C++)
• Develop an Extensible Data Mining Sever (analogous to a database management system)
  

• Completed first prototype of our general-purpose Data Mining Template Library (DMTL). 
• The DMTL library currently supports frequent pattern mining algorithms like Itemsets, Sequences, Trees and Graphs.
• DMTL supports persistency of mined patterns and temporary results.