(Wednesday November 19, 1997 12:30-2:00 PDT 405 Soda Hall)
Paper is still preferred to digital document systems for tasks involving annotating, folding, juxtaposing or otherwise treating the document as a tactile object. Based on the Multivalent Documents model, Multivalent annotations brings to the Web a distributed annotation system with extensive capability to manipulate documents. The individual can use it to customize one's information space, the professor can use it to mark up student papers without the need to print them out, and a distributed community can use it to maintain a set of commentaries on a seminal paper.
The model is extensible not just with new content but with new capabilities as implemented by dynamically loaded Java code, called behaviors, that are as equally sharable as the annotations themselves. This program code has access to all aspects of the document model. Thus, beyond the standard PostIt node and scribbled comment common to numerous annotation systems, in the Multivalent annotations system an ordinary behavior, with no special accomodation in the model, can implement executable copy editor marks: given a suggestion in page annotation to insert a sentence and delete a few words of another, two mouse clicks can materially incorporated those suggestions into the document. Behaviors are written against an abstract document tree so that the same behavior will work on not just HTML but scanned page images and potentially anything in between, as well as other media types such as the implemented video with subtitling. A principal strength of the model is its composition of disparate behaviors, which can and has lead to novel, useful combinations.
Talk Slides [ ppt ]