Computer Science, Electrical Engineering
University of Southern California
Research Interests: OLAP is an elegant approach to efficiently support analytical queries on massive multidimensional datasets. Several fundamental classes of OLAP queries, such as aggregate queries, slice-and-dice queries, or roll-up queries, have been addressed by numerous researchers for the last couple of years. Most of these methods share the disadvantage of either providing only approximate answers by compressing the data or sacrificing the update cost for better query processing performance. However, we propose to employ wavelet transform to efficiently process such queries with exact results and without significant increases on update performance. Leveraging from multi-resolution property of wavelets, we also incorporate approximate query processing in case of space or time limitation as our approach is fundamentally progressive. Furthermore, we prepare and maintain wavelet data by providing a framework to efficiently create, insert, and update data. By developing a real system with its deployment in practice, we address all the essential steps toward realization of our proposed work. We are currently extending our framework by supporting a new class of analytical queries, plot queries, as the most frequently performed queries in scientific applications for extraction of useful information.