Synopses for Massive Data: Samples, Histograms, Wavelets, Sketches
Foundations and Trends in Databases
| Hi-index | 0.00 |
Histogram techniques are widely used in commercialdatabase management systems for an estimation of queryresults. Recently, they have been also used in approximately processing database queries, especially aggregationqueries. Existing research results in this area have beenmainly focused on constructing a histogram to approximately represent, as accurate as possible on an intuitivebase, the original data frequencies. In this paper, we willpropose a novel histogram construction method aiming tominimize the average approximate aggregatio errors; andwe have developed an efficient algorithm to construct nearoptimal histograms to achieve this goal. Our experimentresults showed that the new histogram construction techniques lead to more accurate results than those by existing histogram techniques, and also out-perform the existingwavelet techniques.