Estimating the Selectivity of XML Path Expression with Predicates by Histograms

Selectivity estimation of path expressions in querying XML data plays an important role in query optimization. A path expression may contain multiple branches with predicates, each of which having its impact on the selectivity of the entire query. In this paper, we propose a novel method based on 2-dimensional value histograms to estimate the selectivity of path expressions embedded with predicates. The value histograms capture the correlation between the structures and the values in the XML data. We define a set of operations on the value histograms as well as on the traditional histograms that capture nodes positional distribution. We then construct a cost tree based on such operations. The selectivity of any node (or branch) in a path expression can be estimated by executing the cost tree. Compared with previous methods (which ignore value distribution) our method offers much better estimation accuracy.