curl: Automated Linking of Pages in a Web Document

Andrew Davison
Technical Report 96/23

May 1996

Abstract

Large Web documents cause problems for their users, and for their authors. The most common user problem is getting lost, due to the convoluted organisation of the document, and poor navigational and orientation support tools. The author's main problem is deciding on a suitable organisation, which is expressive enough while not creating an impossible burden of building and maintaining links.

These problems are addressed by curl, a tool for automatically creating the links between the pages of a Web document. curl utilises the metaphor of a book to organise pages, which is both powerful and familiar. Typically, every page is linked to its nearest siblings, its parent, and the top page of the document. Pages can contain two kinds of contents lists, and there are special pages for lists of links to keywords, figures, tables, and new pages. There is also a search engine available from every page.

curl is still being developed (version 2 is the current release), and so this article also contains a discussion of features which will be added to curl in the near future. More speculative extensions are also briefly described.

The complete technical report can be downloaded; it is a 263K gzip compressed Postscript file.

To the curl announcement.