Space-Efficient Indexing of XML Documents for Content-Only Retrieval

More and more documents are stored in semistructured formats like XML. In contrast to traditional information retrieval, the documents can become quite large, and it is often desirable to retrieve not complete documents, but isolated elements that satisfy an information need. To make this possible, the index structures from traditional information retrieval must be adapted to semistructured documents (specifically XML) so that term occurrences can be pinpointed to specific elements inside the documents. This paper explores several enhancements of the index structures and evaluates the advantages and drawbacks of the different versions with respect to index size and retrieval time.