Abstract: The data management system has become the focus of peer-to-pecr computing. Semantic heterogeneity is the first problem to P2P data management system In order to solve this problem,in each data source node,some keywords of table names and attribute names were defined as the medium of semantic mapping. The semantic mapping was automatically created between the heterogeneous data source nodes, which had the same keywords. The keywords became the external schema of the shared tables. But the keywords were not really materialized to the views. All the keywords were distributed to the P2P network using the external schema description files. When a query was executed, once the external schema description files were found,all the aliases would be immediately changed into real table names or attribute names. Using the way, the amount of data backup was reduced, the search algorithm was simplified and the system efficiency was improved.

Key words: Peer-to-peer computing, Data management, Semantic heterogeneity, Data caching, Locating data table