Extended data-flow architecture for data analysis and visualization

Abstract

Since traditional implementations of a data-flow execution model are limited when applied to problems of realistic complexity, several extensions to this model have been made. By using the extensions, the difficulties of traditional implementation are resolved while preserving the virtues of 'pure data flow.' Extensions such as graph evaluation, conditional execution and caching have been embodied in the IBM Visualization Data Explorer software package. Currently, efforts are going on to enhance the implementation of the execution model in Data Explorer in response to user requirements for data analysis and visualization.