Abstract:

At present,most methods of querying the continuous probabilistic XML are discretized.They are not very efficient because the query operators have to process a large number of histogram segments during the query execution.A continuous probabilistic XML query processing technology based on the pdocument model is proposed.Firstly，the pdocument model is expanded to support any continuous distribution by cont node,and the probability density functions and their parameters are encoded in cont node. Secondly, the path that meet user's requirements is found by using the twig pattern match,and then whether a probability query should be executed is decided by using the symbolic form,histograms or using integrals according to the type of continuous distributions to be queried. Standard continuous distributions use the parameters of the symbolic representation in conjunction with some sophisticated functions to compute a query answer,nonstandard continuous distributions that meet integral condition adopt the integral method,and other distributions use the histograms approximating. Experimental results show that this approach has a higher efficiency on both accuracy and response time than the existing approach.

Key words: p-document model;probabilistic XML;continuous distribution;query process