ERCIM News No.20 - January 1995 - GMD

HyperstorM - Object-oriented Database Technology for Innovative Publishing Applications

by Karl Aberer and Klemens Böhm

The objective of the HyperstorM project (Hypermedia Dokument Storage and Modeling) is to investigate the use of advanced object-oriented data base management systems (DBMS) technology for the storage needs in hypermedia publishing applications. GMD-scientists focus on the different systems and concepts , e.g. the cooperative hypertext authoring tool SEPIA, the versioning concept CoVeR and the electronic online journal MultiMedia Forum.

The object-oriented DBMS prototype VODAK is used as a testbed. A document database (D-STREAT) open to the different applications is developed. To this end the utilization of standards like SGML and HyTime is considered. Particular attention is paid to the integration of information retrieval functionality into the hypermedia documentbase. We prove the appropriateness of VODAK as database platform for publishing systems and identify new requirements for the VODAK system. From another perspective, an objective is to support the integration of the systems developed at GMD-Institute for Integrated Publication and Information Systems.

MultiMedia Forum documents are SGML documents: SGML is a language to define document types. With a SGML document-type definition the logical structure of documents of a type is specified. The documents's logical components, the elements, basically make up a hierarchical structure. In a document the elements can be identified by means of so-called markup. In this context, the fact that documents with markup contain more information than layouted ones is important.

Database technology in this context does not only facilitate multi-authoring, but also declarative access and the application of sophisticated versioning concepts. The object-oriented database-management system VODAK, a prototype that has been developed at GMD over the past few years, is the basis for an application to store SGML-documents of arbitrary type. In this context it seems advantageous to model element types from the document-type definition as classes, so-called element-type classes. However, as the set of possible element types is not fixed, these classes need be created dynamically.

The figure illustrates the steps occuring when a document is inserted into the database: First, the document-type definition is transformed - on a syntactical level - into an SGML document, which is conformant to a so-called super-document-type definition. Then inserting document-type definitions into the database (step1) does not differ from the insertion of normal documents, and the generation of element-type classes (step 2) is an internal operation.

In more detail, the MultiMedia Forum is an information service with an editorial team and a worldwide reader's community. From a table of contents, the reader himself selects the items of news he wants to consume. Multi-user mode is possible: the editorial team may carry out their tasks at the same time the readers of the MultiMedia Forum are active. The information in the MultiMedia Forum is not only textual form. Rather there are also multimedia datatypes.

Please contact:
Karl Aberer - GMD
Tel: +49 6151 869 935
or Klemens Böhm - GMD
Tel: +49 6151 869 963