Normal forms for XML documents

Abstract

This paper studies the normalization problem of XML documents with DTDs as their schemas. XML documents may contain redundant information due to a bad designed DTD which implies the similar anomaly dependencies among elements and attributes just as in relational database schema. The concepts of partial functional dependency and transitive functional dependency for XML DTD are proposed. And three XML normal forms: 1XNF, 2XNF and 3XNF, are defined based on the concepts of partial and transitive functional dependency. The concept of lossless join decomposition for DTD is defined by the relational representation of DTD. Finally, two lossless join decomposition algorithms are given for transforming a DTD into 2XNF and 3XNF, respectively. © 2004 Elsevier B.V. All rights reserved.