In: S. Diehl (Ed.): Software Visualization, International Seminar, Dagstuhl Castle, Germany, May 20-25, 2001. Revised Papers, pages 1-73pp. Springer Verlag, LNCS 2269, April 2002.
Abstract: Successful algorithm visualization inherently provides a high level of abstraction, supplying an extra information on the semantics that is behind the code. The postulate of designing highly abstract visualizations is stated as being in a deep contradiction with postulate of automation of the designer's work. The goal of the presented work is partial reconciling these two contradicted postulates. Some elements that significantly increase the level of abstraction may be introduced to the visualization in a strictly automatic mode. To obtain this result, an original method of algorithm animation based on data flow tracing is proposed. Its key idea is to acquire information by observing elementary operations of data flow. For dynamic analysis of non-local flows Petri net formalism is used. The new method has been successfully applied in an algorithm animation system Daphnis.