18. CYBERMAP - Implementation Issues
This chapter describes a series of prototype implementations of CYBERMAP. The initial implementation was done on a Macintosh using HyperCard [Win90]. The first prototype worked only for HyperCard Stacks, but it could easily be generalized as it is straightforward to include plain text documents into HyperCard stacks. The application structure has been kept free from any hard- or software dependent Macintosh or HyperCard particularities and has already been ported to different other environments. Table I.5 resumes the various CYBERMAP ports that have been done in the last five years.
Table I.5 Overview of different CYBERMAP implementations
We implemented the first Macintosh-only version according to the system architecture as outlined in the previous chapter. The main problem with the Macintosh HyperCard implementation was its speed. The first CYBERMAP prototype on a Macintosh IIfx or Quadra took about 30 minutes to cluster moderately sized documents (up to 100 nodes with a few hundred words each). The Macintosh system thus only works as a proof of concept, and the amount of data processing required to generate an overview map for even a small document precludes its use as an interactive tool. It was clear that work on larger, more substantive documents, would require a more powerful system. The distributed and repetitive nature of the task, as well as the quantity of data, called for a parallel computer.