Abstract: A relational database can be viewed as a directed graph constructed by tuples and foreign key references. To facilitate data replication and sharing,semanticrclativity and logically independence should be satisfied when relational data is extracted. Multi-tree structures are employed as clusters of such data extracted from a relational database in this paper. Then the corresponding data extraction approach was proposed and implemented. We evaluated the extraction algorithm on a TPC-C database in Oracle, demonstrating the effectiveness and generalization of the approach.

Key words: Multi tree, Semantic-related, Data extraction