16.3 The Ideal CYBERMAP

The ideal CYBERMAP[13] pops up every time users want to read a hyperdocument. It accompanies them during the whole reading process. It can serve as a starting point by giving readers a graphical outline of the document, but it also shows them all the time where they are in the document by reflecting their navigational movements.

Figure I.86 Schematic view of the use of CYBERMAP

Figure I.86 shows a schematic sample screen of the use of CYBERMAP. The user reads a textual document[14] in the lower left corner of the screen. The CYBERMAP window in the upper right corner of the screen shows a graphic overview map of the document. The CYBERMAP window serves as the primary navigation aid. The fundamental idea is that the original hypertext[15] document remains unmodified, while the CYBERMAP shows an organized view of the data. Figure I.87 displays a screen dump of the actual system where the user is about to browse a document about dinosaurs. The CYBERMAP window is located in the upper left corner.

Figure I.87 Screen dump of Macintosh CYBERMAP implementation

The purpose of CYBERMAP is threefold:

1. It gives an overview of the information contained in the document.
2. Users see where they are in the document. In figure I.86 the actual location of the user is marked by the black rectangle in the middle of the CYBERMAP window.
3. Users see where to go next by perceiving which logical links are connected to their actual location.

One rectangle in the CYBERMAP window represents not only one node of the hyperdocument, but a collection of related nodes. The CYBERMAP representation of such a collection of related nodes is called a hyperdrawer. The hyperdrawer concept offers a means of partitioning the hyperdocument into a set of smaller, manageable fragments. This partitioning can either be disjoint or overlapping: one node may belong to exactly one or to various hyperdrawers. In our work we restrict the distribution of nodes into hyperdrawers to the simplest case: every node belongs to exactly one hyperdrawer[16]. Therefore we get a disjoint partitioning of nodes into hyperdrawers (figure I.88). The hyperdrawers are functionally equivalent to the notion of clusters in information retrieval. We coined a new term to add the notion of sequentiality inside the cluster, because the nodes in a hyperdrawer have to be ordered meaningfully to offer some sort of guided tour through the hyperdrawer.

Figure I.88 Partitioning of nodes into hyperdrawers

The algorithm for the automatic generation of the CYBERMAP is based on fish eye view filtering. CYBERMAP can be used for the production of hyperdocuments written from scratch, but it is more useful for adding similarity links to existing hyper- and non-hyper-documents. As has been mentioned by various authors [Fou90] [Irl90] [Ber90], the transformation of linear documents into hypertextual form, and particularly the automatic generation of links for related nodes poses a serious problem. It is not the goal of this book to describe yet another approach for the automatic generation of links. A simple shallow approach which resembles the algorithm used in the "link apprentice" (see chapter 6.3) is used to find the structure of the hyperdocument.