The optimization of XML queries requires an accurate and compact structure to capture the characteristics of the underlying data. A compact structure works well when the data is uniformly distributed and has many common paths. However, more detailed information needs to be maintained when the data is skewed. This work presents a histogram-based structure to capture the distribution of skewed XML data. It builds upon a statistical method to estimate the result size of XML queries. Experiment results indicate that the proposed method leads to a more accurate estimation.
展开▼
