Indexing XML Objects with Ordered Schema Trees

: XML DBMSs require new indexing techniques to efficiently process structural search and full-text search as integrated in XQuery. Much research has been done for indexing XML documents. In this paper we first survey some of them and suggest a classification scheme. It appears that most techniques are indexing on paths in XML documents and maintain a separated index on values. In some cases, the two indexes are merged and/or tags are encoded. We propose a new method called SIOUX that index XML values on ordered trees, i.e., two documents are in the same equivalence class is they have the same schema tree, with identical elements in order and number. The method can be seen as an extension of relational DBMS systems, where schemas are separated from data, and data are manipulated with their associated metadata. We use a simple benchmark to compare our method with two well-known European products. The results show that indexing on full trees is better for pure structural search in most ca...

• Type: Conference
• Year: 2004
• Where: BDA
• Authors: Laurent Yeh, Georges Gardarin
• Keywords: BDA 2004, BDA 2007, Indexing Xml Documents, Structural Search, XML Documents