This paper describes the development of a hypertext system for an engineering encyclopedia, called the Compendium, on a CD-ROM running on a personal computer. The printed version of the original text is 3000 pages of textual and graphical information, and the document is well structured and well organized in its manual form. A system was developed to provide computer-based reading enhancements; this project was a full-scale research and development exercise rather than a prototype development. The resulting system enables the reader to access material by searching through the original table of contents, by using a back-of-the-book index, by executing a boolean search on the text, or by browsing. The browser is a hierarchical browsing facility that allows “exploration by progressive display of detail” and thus “allows the user to exploit the extensive organization of the Compendium in locating information.”

The original textual form of the Compendium was converted into hypertext format automatically. The first phase of this conversion was to identify text units or hypertext nodes. The nodes were identified using structural information from the original document, which yielded 1100 hypertext units corresponding to the original encyclopedia entries. Phase 2 of the conversion was to identify the links between these text units, and the developers found that each of the original entries has about 10 explicit cross references to other entries as well as some implied cross references. Only the explicit references were used as hypertext links in the system.

So where was the difficulty? From this paper it seems there was none. It looks like the whole process was a straightforward text transformation operation. In the “lessons learned” section of the paper, the author points out that a lot of time was spent becoming familiar with the original document, but the paper does not allude to any significant or particularly useful lessons gleaned from this exercise. Overall, this paper is lightweight in terms of content and length, presumably because it is a conference paper. Also, the title suggests the paper is going to discuss the text transformation operation, but it does not. A more suitable title would have been something like “An outline of a hypertext-like system developed for searching an engineering encyclopedia.” My overall impression of the paper is that something more interesting went on in the project, but it is not reported here. On the other hand, a hypertext system with 1100 nodes, 20 Mb of text, and 150 Mb of graphics is a big system, which makes this project important.