Generalized SGML repositories: requirements and modelling

In numerous fields of technical documentation and publishing, paper media are giving way to electronic media. The SGML standard is gaining increasing importance. The documents processed, in particular in the aircraft industry but also in other fields like legal publishing, are generally of large size, highly structured and regularly updated. Their production to the SGML format creates a need in terms of document management systems. A repository architecture, supporting different kinds of activities such as consultation or edition, is a requirement for such a production. We give an overview of the requirements for such a repository, starting with those already widely recognized and used as a basis for on-going experiments and products. We then detail some unforeseen requirements that are just emerging. Finally, we will analyse the impact of these requirements on information modelling inside and outside the repository. The analysis of requirements reveals some shortcomings of the SGML exchange model itself. An evolution towards the HyTime standard enables most of these shortcomings to be offset. For the repository internal model, we propose an object-oriented data model that meets most of the requirements.