Developing Presentation Based on Diagrammatic Information Structure by a Virtual Human Agent

本研究项目的目的是设计和开发一个系统，其中特定数据或信息的本质、结构和重要或显著点由用户指定并/或自动提取，然后生成角色动画来解释它们。2005年，我们开发了一个系统，用户指定表格中元素之间的关系，然后生成动画。这些关系以树形结构表示。为了产生动画，遍历树，并将访问的元素和关系解释为包括语音文本和演示者手势的呈现场景。解释规则用模板表示，2006年，我们首次将该系统扩展为可以生成图形和表格的演示动画的系统。在图形表示中，代理可以解释值序列的趋势(例如，增加或减少)以及显著的点(例如，最大值或最小值)。接下来，我们研究了根据显著点的特征自动选择候选点的机制。它允许用户选择一些候选对象，然后自动生成演示动画，解释所选择的显著点。上述工作在第五届IASTED可视化、成像和图像处理国际会议(VIIP2005)和SIGGRAPH 2006研究海报中介绍。

The purpose of this research project is to design and develop a system where the essence, structure, and important or remarkable points of particular data or information are specified by a user and/or extracted automatically, and then a character animation is produced to explain them.In 2005, we developed a system where a user specifies relationships among elements in tables, and then produce an animation. The relationships are represented in a tree structure. To produce an animation, the tree is traversed and the visited elements and relations are interpreted into presentation scenarios including speech texts and presenter's gestures. The interpretation rules are represented by templates.In 2006, we first extended the system into one that can generate presentation animation for graphs as well as tables. In graphical presentations, the agent can explain the tendencies of sequences of values (for example, increasing or decreasing) as well as the remarkable points (for example, maximum or minimum). Next we studied the mechanism which automatically selects candidates of remarkable points with their features. It enables users to choose some of the candidates, and then the presentation animation explaining the chosen remarkable points is produced automatically.The above work is presented in 5th IASTED International Conference on Visualization, Imaging (VIIP2005), and Image Processing and SIGGRAPH 2006 Research Posters.