III: Small: Collaborative Research: ANTE - A Four-Tier Framework to Boost Visual Literacy for High Dimensional Data

这是一个合作研究项目，涉及纽约州立大学石溪分校和本科院校联合学院。随着数据的海量可用，理解和熟悉数据的需求变得越来越重要。现在对拥有从数据中提取意义的技能的个人的需求很大。许多机构都开设了数据科学的学术课程，但对于很大一部分人来说，回到学校正式学习这个主题是不可能的。此外，并不是每个人都真的需要成为一名正式的数据科学家，才能在这个日益以数据为中心的社会和工作场所具有竞争力，在这些社会和工作场所，变得更懂数据可能会有很大好处。可视化，如大多数人熟悉的条形图、折线图、地图等，有助于解释数据。然而，今天的数据集往往结合了许多不同类型的信息，因此太复杂了，无法用这些基本的可视化来表示。这个项目的目标是开发一个可视化系统，它可以用一种用户无需大量培训就能理解复杂数据的方式来表示数据。这将涉及可视化技术的进步以及以引人入胜的方式呈现可视化的新方法。在这个项目中开发的ANTE(呼吁、叙述、转变、参与)系统具有很好的潜力，有助于提高公民在日益以数据为中心的社会中成为更有知识的参与者的能力。该项目为纽约州立大学石溪分校的研究生和联合学院的本科生提供研究培训。可视化工具和游戏将为所有教育级别的数据和视觉素养教学创造一个极好的环境。ANTE框架寻求通过开发新的解决方案来实现其目标，这些解决方案涉及以下四个相辅相成的要素：呼吁、叙述、转变、参与。ANTE将通过定义新的强大技术来吸引用户现有的视觉素养，这些技术可以忠实地将复杂的高维数据转换为更简单的表示形式。一种这样的表示是一种新颖的2D上下文数据映射，与其他这类映射不同，它可以维护数据矩阵中的所有关系--数据到数据、数据到属性、以及属性到属性。另一种是交互式3D阴影显示，它取代了标准用于高维可视化的复杂散点图矩阵。Ante将使用自然语言来描述它产生的可视化效果。Ante还将使用动画来展示数据集的不同表示如何相互转换。该项目将调查此类动画的使用，以教授用户如何解释更高级的可视化效果。最后，ANTE通过以下方式吸引用户参与数据：(1)通过视觉因果分析的方式支持故事创作；(2)通过将数据与从网络规模的图像数据库中检索的上下文图像融合来设计引人注目的信息图表的能力；(3)使用类比学习来教用户从熟悉的可视化中学习更复杂的可视化的叙事界面；以及(4)使用从游戏化中收集的技术来激励用户参与我们所有建议的技术的用户评价研究的框架。更多信息、研究论文和开发的人工制品，如可视化系统的网络界面、数据、视频链接等，可在项目网站(http://www3.cs.stonybrook.edu/~mueller/ANTE/).上获得

This is a collaborative research project involving SUNY Stony Brook and Union College, an undergraduate institution. With the massive availability of data, the need to understand and be comfortable with data has gained increasing importance. There is now a great demand for individuals that have the skills to extract meaning from data. Academic programs in data science been created in many institutions, but going back to school to formally study this topic is not possible for a large segment of the population. In addition, not everyone really needs to become a formal data scientist to be competitive in this increasingly data-centric society and workplace where it can be of great benefit to become more data literate. Visualizations, such as the bar charts, line plots, maps, etc. that most people are familiar with, are helpful in explaining data. However, today's data sets often combine many different kinds of information and are, therefore, too complex to be represented with these basic visualizations. The goal of this project is to develop a visualization system that can represent data in such a way that a user can make sense of complex data without extensive training. This will involve advances in visualization techniques as well as novel approaches to presenting visualizations in an engaging way. The ANTE (Appeal, Narrate, Transform, Engage) system developed in this project has good potential to help increase the ability of citizens to become more knowledgeable participants in an increasingly data-centric society. The project provides research training for graduate students at SUNY Stony Brook and undergraduate students at Union College. The visualization tools and games will make an excellent environment for teaching both data and visual literacy, at all education levels. The ANTE framework seeks to achieve its goal by developing novel solutions that address these four complementing elements: Appeal, Narrate, Transform, Engage. ANTE will appeal to the user's existing visual literacy by defining new powerful techniques that can faithfully transform complex high-dimensional data into simpler representations. One such representation is a novel 2D contextual data map that unlike other maps of this kind can maintain all relationships in the data matrix -- data to data, data to attribute, and attribute to attribute. Another is an interactive 3D shaded display that replaces the complex scatterplot matrices that are in standard use for high-D visualization. ANTE will use natural language to narrate the visualizations it produces. ANTE will also use animations to show how different representations of the data set can be transformed into each other. The project will investigate the use of such animations for teaching users to interpret more advanced visualizations. Finally, ANTE entices user engagement into data by offering (1) support for story authoring by ways of visual causality analysis; (2) capabilities for designing compelling infographics by fusing data with contextual images retrieved with web-scale image databases; (3) a narrative interface that uses learning from analogy to teach users more complex visualizations from familiar ones; and (4) a framework that employs techniques gleaned from gamification to incentivize engagement in user evaluation studies for all of our proposed techniques. Further information, research paper and developed artifacts, such as web interfaces to the visualization systems, data, video links, etc. are available at the project web site (http://www3.cs.stonybrook.edu/~mueller/ANTE/).