Ranking for Approximated XQuery Full-Text Queries

摘要

Treating structural conditions included in an XQuery Full-Text expression as desiderata rather than mandatory constraints could be beneficial for better answering user's informational needs. Using an approximated semantics requires however a mechanism for calculating a score for each resulting tree, and an algorithm for efficiently finding the best results. Both these issues are explored in this paper. We present algorithms for query evaluation that uses a threshold approach in order to improve performance; the algorithms rely on some properties of the functions used to calculate the score. Moreover, we propose a method, based on the new concepts of Path Edit Distance and Comparison Satisfaction Ratio, for calculating the structural score of a tree that partially satisfies conditions over tree structure and element values.