Applying a Data Miner To Heterogeneous Schema Integration

Anapplication of data miningtechniques to heterogeneous database schemaintegration is introduced. We use attribute-oriented induction to minefor characteristic and classification rules about individual attributes from heterogeneous databases. Eachmining requestis conditionedona subsetof attributes identified as "common"betweenthe multiple databases. We developa methodto comparethe rules for twoor more attributes (fromdifferent databases)anduse the similarity betweenthe rules as a basis to suggestsimilarity betweenattributes. As a result, weuse relationships betweenandamongentire sets of attributes frommultiple databasesto drive the schemaintegration process. Ourinitial efforts and prototypes applyingdata mining to assist schemaintegration provepromisingand, wefeel, identify a fruitful application area for data miningresearch. goywords: schemaintegration, multi-database interrelationships, attribute similarity, data mining, attribute-oriented induction.

• Type: Conference
• Year: 1995
• Where: KDD
• Authors: Son Dao, Brad Perry
• Keywords: Attribute-oriented Induction, Data Mining, Heterogeneous Database Schemaintegration, Heterogeneous Databases, KDD 1995