Research And Implementation Of Methods Of Query Transformation For Information Integration

The task of information integration system is to integrate multiple heterogeneous data sources to provide users with a unified view. It makes users feel those data from the same data source, and enables users to access data sources conveniently. It hides the differences from underlying sources, and decides the content which the system allows users to access. When users ask for a query, the system needs to assign the query to each source to execute. Query transformation works for the process.In this paper, we researches the problems of query transformation based on the Wrapper/Mediator architecture for the integration system of heterogeneous data sources which consider the GAV approach as mapping between global schema and sources. Query transformation involves two processes:transforming a mapping to internal expression; unfolding and rewriting user query. There are serval problem in it:how to express the structure of mapping to make the extension of query more convenient; how to rewrite and optimize the query to get equivalent and efficient plan.This paper improves the structure of query tree of relational algebra according to these problems, and proposes RATree to express the mapping and query tree. User query can be unfolded by the RATree, and be equivalently transformed into the query on local schema; We pose the derived rules of Equivalence Transform of Relational Algebra, and demonstrate the efficiency of rules; We pose query decomposition method and query rewriting method based on RATree. They can reduce the cost of intermediate results and the complexity of the query plans, improve concurrency of queries. We design the application of query transformation, and give an example to illustrate and validate the query transformation method.