Authors of highly formalized documents like technical documentation or scientific texts have to manage references of citations, bibliographical and URL information, product names and trademarks or application specific glossaries. A distributed system could store meta information about this information on the server and allow the author to automatically retrieve references to contents of this repository.
For example, the author refers in his document to the Microsoft Windows® Operating System. The reference management system will find out, that the term Microsoft Windows® is a reserved trademark of the Microsoft Corporation and will mark up the term and add a corresponding legal notice to the document.
As another example there may be a glossary of product names and identifiers that authors have to use to adhere to the company's style guide. If the reference management identifies the use of an undesired word, it will indicate this misuse to the author and suggest a replacement.
For example: replace “band width” with “bandwidth” (one word).
There should be a web interface to the reference management system to add, delete or modify reference definitions.
It would be ideal to use an XML-database for storing the contents of the repository and using technologies like XPath or XQuery to request contents from the database. A pipeline-based XML publishing system like Apache Cocoon could be used to create a web-interface to the repository. As most data structures in the context of a XML based documentation are XML documents and document fragment, a framework like Cocoon could offer decent support to creating web interfaces.