Seamless Integration of Diverse Data Types into Exploratory Visualization Systems

摘要

Visual data exploration involves presenting data in some graphical form and allowing the human to interactively gain insight into the data. Visual data exploration techniques have proven to be very useful for exploratory data analysis, especially for mining large databases. However, the lack of explicit content semantics within the framework of the visualization tool makes interpretation of the visualizations difficult at times. The aim of our research is the development of a comprehensive framework for integrating semantics of data into an existing visualization tool in a flexible yet minimally intrusive manner. The first stage of the framework involves a transformation process that enables the mapping between a wide range of data types to a form the visualization system can process. This stage exploits the strengths of XML for handling semantics extraction, flexible data modelling of domains and managing the data mapping rules to conform the data to the requirements of the visualization tool. The second stage of the framework enables the visualization tool to access and present the semantic information separate from the data display. We demonstrate and validate this framework by illustrating the extension of XmdvTool, a multivariate data visualization system, to handle varied data types such as categorical fields and unstructured text files. We believe the framework is sufficiently powerful to incorporate other forms of data, as well as to be used to extend other visualization tools.