Building Hypertexts under the Influence of Topology Metrics

Abstract

One of the greatest problems inhibiting the spread of hypertext is the cognitive overload on authors. In addition to authoring the content of each node, an author is responsible for creating information links and link anchors while also having to maintain an overall balanced hypertext document. Our work has addressed this aspect of the authoring problem by offering an author a menu of potential information links calculated by computing node-node similarities. We are also investigating whether the values of graph topology metrics can be used as an aid to a hypertext author in maintaining an overall balanced document by measuring the overall topology of the hypertext being generated and then, as new links are created, re-measuring the overall topology to determine how the newly created links effect the overall hypertext layout. Automatic link creation and our own proposal for topology measurement are not exclusive and in previous work we have combined them together in automatic text to hypertext conversion.