Semantic Management of Deduplicate Tuples in the Relational Databases

Relational database is a collection of relations. Duplicate tuple existence is common in many real time relational databases. In a relational database, if the same real-world entity is represented by more than one tuple, then such tuples are called duplicate tuples. Finding duplicate tuples and then replacing them by one best tuple is called a fusion operation. Whenever duplicate tuples are found in the relations of any database, those tuples must be replaced with one special best approximate tuple that represents the joint information of all the duplicate tuples. Present study proposes new techniques to find duplicate tuples and then remove those duplicate tuples with the correct real world tuples. In the first step duplicate tuples in the relation are classified based on the class label and in the second step then for each set of duplicate tuples functional dependency method or union method is applied to replace duplicate tuples with the corresponding correct real world single tuple. One possibility is to replace one set of duplicate tuples with one correct real world tuple. Another possibility is to replace two or more sets of duplicate tuples in the relation by one set of correct real world tuples. Sometimes the removal of duplicate tuples in the relations of any relational database can create referential integrity violations. All such violations must be controlled and coordinated syntactically as well as semantically in relations.