{"id":61,"date":"2017-05-04T11:18:49","date_gmt":"2017-05-04T15:18:49","guid":{"rendered":"http:\/\/sites.nd.edu\/chaoli-wang\/?page_id=61"},"modified":"2025-08-30T10:00:45","modified_gmt":"2025-08-30T14:00:45","slug":"mining-transition-graphs","status":"publish","type":"page","link":"https:\/\/sites.nd.edu\/chaoli-wang\/code\/mining-transition-graphs\/","title":{"rendered":"Mining Transition Graphs"},"content":{"rendered":"

Introduction<\/h2>\n

Our graph-based approach to time-varying volume data analysis and visualization derives a transition graph from the time-varying data. In the transition graph, a node denotes a state which represents a group of spatiotemporally neighboring blocks, and a directed edge between two states indicates their transition probability. The graph mining solution then automatically extracts meaningful features from the transition graph. This is achieved through the utilization of a series of graph analysis techniques including graph simplification, community detection, and visual recommendation. Our goal is to allow significant reduction of cognition overhead and interaction cost when exploring the underlying time-varying volumetric data using the graph-based technique.<\/p>\n


\nA community is detected from the transition graph and highlighted using Bubble Set in the graph view (left). The corresponding blocks of the community are rendered in the volume view (right).<\/p>\n

Download<\/h2>\n

We provide the source code of the important steps in our approach. These important steps include state clustering, graph simplification, SimRank, and graph distance calculation. The source code has been compiled successfully under Ubuntu 14.04. The source code makes use of the following three external libraries:<\/p>\n