Visualization of Splitting and Merging Processes

Information about objects that split or merge is often generated dynamically as a by-product of computation or in the observation of real-world behavior. Visualization tools for such processes must not only reveal temporal patterns and spatial organization but should also accommodate on-the-fly generation of split-merge information. This paper develops a formal structure for split-merge processes that provides a suitable underpinning for dynamic visualization tools. The structure allows split-merge processes to be constructed dynamically and supports operations such as concatenation and partitioning while maintaining the underlying split-merge structure. The paper also proposes a method for abstracting split-merge substructure to assist in the visualization of large systems. The abstractions can be used to produce visual simplification and to guide layout heuristics in producing better visualizations. The implications of visualizing split-merge processes with time lines are explored. Auxiliary visualization strategies based on dynamic boxes and tree rings are introduced to enhance information content of focus areas in larger time-line visualizations. The concepts are illustrated using data from timings of a network protocol and from the detection of patterns in scientific video data.