Browsing Semantics in the Tower Model

In many hypertext systems the result of the follow-link operation is that the "current node" is replaced by a new node. Only one node is accessible at any time. However, systems like Guide and most WWW browsers can display several nodes at once, and you can click on anchors in either of these nodes to follow links that may replace nodes or add more nodes to the ones already displayed. This is modeled in the Tower Model by allowing that the source and destination anchors of links are connected to (or part of) several nodes.

In order to generalize the notion of "current node" to such situations we define the State of a hypertext as a boolean function S on the objects of the hyperdocument. The value true means that the object is accessible (or visible, or current), while false means that the object is not accessible.

We only consider accessibility for single node-views, i.e. for individual views (node tower objects) in a city. When a node is accessible we assume that all the anchors to which it is connected are also accessible.

The basic operation for hypertext: follow link is defined as a function from states to states. A link l = (i, s, d, v) is enabled when S(n) = true for some node n to which the source anchor s is connected. The result of following an enabled link l is a new state S':

This definition says that the destination node(s) become accessible, the source node(s) become inaccessible unless they are also in the destination, and state for the other nodes remain the same.

Using this definition the behavior of a hypertext can be represented by means of Boolean Petri-net. All other operations in the Tower Model are also expressed in the Petri-net formalism. Boolean Petri-nets can easily be translated to standard Petri-nets. Thus, the analysis techniques for Petri-nets can be used to find potential browsing problems with hyperdocuments, such as small loops and dead ends.