Optimization of Data Retrievals in Processing Data Integration Queries

Data integration systems often suffer from performance bottlenecks due to the network overhead incurred by frequent and large-scale data retrievals. In this paper, we propose to develop query reconstruction mechanism in source wrappers to exploit data sharing across multiple data retrievals, and hence to optimize the performance of query execution. We propose a derived-based query reconstructing method in this paper. This method firstly builds lattices to describe potential data sharing opportunities among queries, and then constructs new queries accordingly to exploit them. Furthermore, a corresponding method is present to describe how to compute answers from the reconstructed queries. Finally, Experimental results are described to show the impact of our solution on the performance of processing data integration queries.