17.1 Identification of Hyperdrawers

To partition the nodes into hyperdrawers, three related problems have to be solved:

Problem 1 - Put related nodes into the same hyperdrawer:

The hyperdrawers have to be identified and related nodes have to be grouped together. An easy way to do this for textual nodes is to cluster the nodes by using automatic indexing techniques as described in chapter 1.

Figure I.89 Hyperdrawer hash function F

In order to find an even distribution among the hyperdrawers, we have to identify a hash function F (figure I.89) with the following ideal properties:

• Because of ergonomic and cognitive aspects F should distribute the nodes among the hyperdrawers as equally as possible. Ideally every hyperdrawer should contain the same number of nodes:
  [[universal]] i, k: # elements of hyperdraweri =# elements of hyperdrawerk;
• F should ensure that there are not more than about twenty [17] hyperdrawers, because if there are too many boxes on the screen, it gets too cluttered:
  number of hyperdrawers <= 20;
• F has to group together in one hyperdrawer the most related nodes. This means that the contents of every node have to be analyzed and that a measure for the affinity between nodes has to be defined.

Problem 2 - Name the Hyperdrawers:

Meaningful names for the groups, or hyperdrawers, have to be identified.

Problem 3 - Identify the Topology:

After the hyperdrawers and the partitioning hash function F have been found, the hyperdrawers have to be drawn in the CYBERMAP in as an informative and aesthetically appealing manner as possible. The easiest approach would be to distribute the hyperdrawers at random on the map. Ideally, however, one would like to put more information about the structure of the document into the CYBERMAP. Therefore the document structure has to be identified based on an analysis of the contents of the document. The linking structure of the hyperdrawers has a priori nothing to do with existing links in the hyperdocument.