Efficient probabilistic XML query processing using an extended labeling scheme and a lightweight index

Recently there is a growing interest in the data model and query processing for probabilistic XML data. There are many potential applications of probabilistic data, and the XML data model is suitable to represent hierarchical information and data uncertainty of different levels naturally. However, the previously proposed probabilistic XML data models and query processing techniques separate finding data matches with evaluating the probabilities of results. Therefore, they should repeatedly access the data and need to get full data of paths given in queries to calculate the probabilities of results. In this paper, we propose an extended interval-based labeling scheme for the probabilistic XML data tree and an efficient query processing procedure using the labeling scheme. Against previous researches, our method accesses only the labels of data specified in queries and finds data matches simultaneously with evaluating the probability of each data match. Also, we present an extended probabilistic XML query model with the predicates for the values of probabilities and a lightweight index for those probabilities in order to eliminate unnecessary access to data that will not be included in results. Experimental results show that our approach is efficient in probabilistic XML query processing and our index scheme significantly improves the performance of query processing when the predicates for the values of probabilities are given.