Duplicate detection and deletion in the extended NF2 data model

Abstract

A current research topic in the area of relational databases is the design of systems based on the Non First Normal Form (NF2) data model. One particular development, the so-called extended NF2 data model, even permits structured values like lists and tuples to be included as attributes in relations. It is thus well suited to represent complex objects for non-standard database applications. A DBMS which uses this model, called the Advanced Information Management Prototype, is currently being implemented at the IBM Heidelberg Scientific Center. In this paper we examine the problem of detecting and deleting duplicates within this data model. Several alternative approaches are evaluated and a new method, based on sorting complex objects, is proposed, which is both time- and space-efficient.