Rank queries in probabilistic data integration systems for digital library federation

In digital library federation, a challenge is how to exchange and share information on heterogeneous digital libraries. The mediator-based approach of data integration provide a uniform view by creating the semantic relationship between sources and mediated schema. Since automatic matching produce probabilities, itpsilas necessary to take the probability of mappings into consideration. Further more, for tuplespsila probabilities arise as an additional ranking dimension, both tuple probabilities and scores be factored in the interpretation of top-k queries in probabilistic databases.In this paper, after proposing a architecture of data integration for DLF, we demonstrate the probabilities for the schema mappings and query answerings. Then, having discussed the opt_U-kRank algorithm, we describe the revised_U-kRank algorithm which may improve the efficiency of execution. Finally, based on the experiment results of two U-kRank algorithms, we discuss the effect of probabilities distribution on executions.