CRII: III: Evaluating Provenance Visualizations for the Presentation and Communication of Investigative Data Analysis Processes

点击翻译按钮获取中文摘要

Many types of data analysis involve complex investigations with large data sets, open-ended explorations, and iterative hypothesis testing. The high complexity and potential variability in human analytic processing can make it difficult to remember the steps and rationale that led to the formation of hypotheses, the generation of specific data views, and the realization of conclusions. There is an established need for reviewing analytic processes, and many tools provide visualizations to do so; however, it is not well known how helpful the tools are for practical purposes. In particular, little visualization research has evaluated the effectiveness of visual representations for the purposes of communication and presentation of analysis history. The project will explore new designs for visually representing the history of data analysis and evaluating the effectiveness of different designs. The outcomes of the visualization designs and empirical evaluations will yield direct benefits to professional analysts and scientists by improving the ability to review and communicate analysis records. Effective presentation and communication of analytic processes is essential for understanding the underlying arguments behind decisions, and visual presentation will facilitate review of analysis processes by making it possible to understand analytics strategies and their effectiveness. Reviewing analysis approaches and strategies will allow analysts to identify problems with existing methods, improve those methods, and better train new analysts and scientists.This project will provide the foundation for an extensive effort of the design and evaluation of visualizations for the presentation and communication of the history of an analysis process, which is also known as analytic provenance. Through the development of new evaluation methodology, the project will enable the assessment of how well specific visual elements of provenance visualizations contribute to successful presentation of different types of provenance information. The effort will include detailed collection and coding of samples of provenance data that will be used as a basis for evaluating visualizations. The research will explore novel visualization designs for presenting different types of captured information. In addition, the effort will investigate methods to automatically generate provenance visualization, and it will study the effectiveness of different forms of automated presentations.