We conclude this paper with a few guidelines for encoding document content in a display-independent manner. Electronic encodings that adhere to these guidelines will enable multiple uses of the same electronic source. Though the notion of archiving information in its richest possible form is itself not new, we note that such ideas have been exclusively motivated in the past by the need to display information visually. The richest representation for the specific problem of being able to accurately reproduce the visual appearance of information is not necessarily appropriate for computing/ on the information. Visual presentations, as pointed out earlier, are optimized for human consumption, and therefore necessitate explicit human intervention in performing intelligent/ manipulation of the content.
Our work brings a fresh perspective to this issue by addressing the problem of aurally rendering complex information. It points out that the visual presentation that we are all familiar with, e.g.,the printed version of this paper, is just one possible view of the information content, not the information itself. This insight leads naturally to the approach used in AsTeR , namely the development of high-level information representation and the rendering of such representations in different modalities.
To ensure a multiplicity of uses, the digital library should archive information in its richest form. Such encodings should be capable of producing high-quality renderings in the various output modalities, e.g.,a well-formatted PostScript or PDF file containing high-resolution fonts, audio renderings that exploit the various features of an auditory display, etc. A digital library may choose to archive one or more of the ``display'' forms in addition to the high-level encodings as a means of optimizing information delivery. However, archiving information in any of these ``display'' forms is equivalent to archiving information on printed paper. Hence, such ``display'' representations should not be viewed as a replacement for the high-level encoding.
Retaining the high-level encoding that generates the various renderings will facilitate:
The Chicago Journal of Theoretical Computer Science is an online journal to be published in LaTeX (Url: Hot Link) and fulfills these ideals. The markup recommended to authors has been carefully designed to abstract out all layout details by the Managing Editor, Prof. Mike O'Donnell, and we hope to aurally render articles from the journal using AsTeR . See [O'D93][O'D92] for a description of the work leading up to this project.