by Alan Greg Turransky
Submitted to the Program in Media Arts and Sciences, School of Architecture and Planning on July 15, 1993 in partial fulfillment of the requirements for the degree of Master of Science in Media Arts and Sciences at the Massachusetts Institute of Technology.
The abundance of layout problems commonly associated with the presentation of visual information on computer displays necessitates that computer systems be incorporated with graphic design knowledge to effectively and intuitively aid users in presenting, customizing , and organizing this form of data. Current methods of encoding such knowledge requires that human designers verbally translate their expertise into a set of programmable rules, frames, cases, or constraints. Computer systems which can be trained to learn the techniques designers use to effectively present visual information, by having a designer demonstrate their application on a working example may provide a more natural means of translating this type of knowledge from its original visual form into the electronic environment, without the necessity to first translate it into a textual representation.
This thesis describes a system which uses a machine learning technique called Programming by Demonstration to overcome this translation problem and enable the transformation of visual ideas into usable symbolic forms. It offers a working model, called the Abatan system, for capturing re-usable, graphic design knowledge from interactive user demonstrations.
This work was sponsored in part by the Alenia Corporation, the Joint National Intelligence Development Staff, and the Kansa Corporation. The views and conclusions in this document are those of the author and should not be interpreted as necessarily representing the official policies, expressed or implied, of the Alenia Corporation or the Joint National Intelligence Development Staff.