Extensible Indexing in XML Databases

Extensible indexing aims at providing indexing support for arbitrary queries. Existing XML indexing approaches do not provide extensibility as they are tailored to specific indexing requirements. To facilitate and further improve XML indexing, this paper applies the concepts of extensible indexing to XML databases. So far, extensible indexing has been studied in the realm of object-relational database systems. In that con- text the focus is on indexing user-defined data types and operations via generic index templates and/or extensible index interfaces. Generic index templates (e.g. GiST) reuse one index structure for different data types by adapting type-specific operators. Extensible index interfaces (e.g. Oracle Data Cartridges) enable to seamlessly integrate new index structures into a database. This paper similarly provides extensible index structures and an extensible index framework. Extensible index structures enable to index the content and structure of XML documents with only a small number of index structures. By adapting comparison operators, it is for example possible to index not only values but also paths with a B+-tree. The extensible index framework provides an index model as the basis for processing and integrating these index structures in an XML database.