A Histogram Method for Summarizing Multi-dimensional Probabilistic Data.

Autor: Iqbal, Ashraf, Wang, Hai, Gao, Qigang
Předmět:
Zdroj: Procedia Computer Science; May2013, Vol. 19, p971-976, 6p
Abstrakt: Abstract: Currently, many database applications deal with large imprecise and uncertain datasets. Probabilistic data summarization has recently emerged and has already become an active research area in the database community. In this paper, we propose a data summarization method to summarize multidimensional probabilistic data using histograms. The proposed method iteratively constructs a histogram to represent the probabilistic data while maintaining a trade-off between minimizing the relative entropy among probability distributions and minimizing the space used by the histogram. The experimental results show that the proposed method achieves small errors for various compression ratios. [Copyright &y& Elsevier]
Databáze: Supplemental Index