Separable hyperstructure and delayed link binding
David F. Brailsford · ACM Computing Surveys · 1999
Abstract: As the amount of material on the World Wide Web continues to grow, users are discovering that the Web's embedded, hard-coded, links are difficult to maintain and update. Hyperlinks need a degree of abstraction in the way they are specified together with a sound underlying document structure and the property of separability from the documents they are linking. The case is made by studying the advantages of program/data separation in computer system architectures and also by re-examining some selected hypermedia systems that have already implemented separability. The prospects for introducing more abstract links into future versions of HTML and PDF, via emerging standards such as XPath, XPointer XLink and URN, are briefly discussed. Background Research on electronic hypertext systems stretches back for more than 30 years to Engelbart's [Engelbart 1968] and Nelson's [Nelson 1987a] astonishingly far-sighted work. A body of hypertext transcends normal textual documents not only by allowing user-specific annotations to be added to the corpus, but also by enabling a set of intra- and inter-document hyperlinks to weave documents together into an associative `web'. Indeed, today's World Wide Web can be regarded as a reification-- albeit flawed and full of compromises-- of what these pioneers had in mind. Over the years the idea of linking has evolved from being a purely text-related activity into the wider world of