Dynamic Data Mining Using an Electro-Optical Data Warehouse

Ultra large user-oriented distributed multimedia information systems that manage terabytes of data must provide rich functionality so that interesting new applications can be addressed. Since a wealth of data, information, and knowledge are resident within these vast repositories; a variety of data mining techniques have developed. These techniques are very computationally intensive and require the movement of large amounts of data. In the search for new approaches, optics may be able to help since photons have the very attractive properties of high speed, non-interference, and parallelism. Optical systems can accommodate a large number of parallel, high-bandwidth channels, and optical storage devices have very high storage densities. In this proposed program we will postulate an electro-optical computer architecture and examine the feasibility of executing a number of data mining and knowledge discovery algorithms on ultra large multimedia data/knowledge bases with the express purpose of increasing performance and functionality. Electro-optical special purpose architecture enhancements and advanced adaptable memory design/configurations will be investigated in order to postulate an innovative electro-optical computer architecture. Alphanumeric and image data mining and knowledge discovery algorithms will be selected and evaluated on the electro-optical computer architecture. In addition, the feasibility of interfacing with intelligent agents on the web and with mobile hand-held computing devices will be investigated.