Dealing With (Un)Structuredness in XML Data and Queries Using Relational Databases

An XML database can contain documents with varying degrees of schema information. The queries can also range from fully specified structured SQL like queries to partially specified regular path expression queries. Relational databases are widely used to store and query XML data and various schemes have been proposed to this end. They either use DTDs or assume schemaless data. This paper shows how more advanced schema information (like XMLSchema), even if partially available, can be used effectively to answer queries. The paper also shows how the interaction between the amount of schema information available and the query workload plays an important role in choosing a decomposition strategy into relational tables, suited to that workload.