Abstract (摘要)
While Scientific Visualization techniques are used for the clarification of well-known phenomena, Information Visualization techniques are used for searching for interesting phenomena. There are therefore important differences between the two fields of research. The two fields of research are compared, but since this paper is meant for Scientific Visualization researchers focus will be on explaining Information Visualization techniques. This paper list the main characteristics of the two fields of research, describes common Information Visualization techniques and discusses differences and similarities in the software that commonly is used in the two fields of research.
科学可视化技术用于明确著名的现象，信息可视化技术用于寻找有趣的现象。因而两种研究领域之间存在着重要的区别。本文比较了这两种研究领域，但是由于本文面向的是科学可视化的研究人员，因此关注点集中在信息可视化技术的解释。本文列出了两种研究领域的主要特征，描述了常见的信息可视化技术并且讨论了两种研究领域中常用软件的异同点。
1. Introduction （引言）
Some researchers mainly use visualization for finding interesting phenomenon in completely unknown data, whiles others use visualization for the confirmation or rejection of hypotheses. These two scientific communities live separate lives, with very little sharing of knowledge between them. While there naturally are important differences between the two kinds of visualization, it is the hypothesis of this paper that an understanding of both of these two ways of using visualization potentially could lead to new ways of visualizing data for the benefit of both communities. This article therefore contains a summarization of the most important and irreconcilably differences between the ways of thinking. However, since this article is presented at a Scientific Visualization symposium, the focus will here be on Information Visualization ideas and techniques.
一些研究人员主要使用可视化来从完全未知的数据中寻找有趣的现象，而另外一些研究者使用可视化来验证或者否定假说。这两种科学团体各行其是，相互之间的知识共享很少。两种可视化之间有着重要的区别，本文假设对于这两种使用可视化技术方式的理解可以寻找出让两种团体都获益的新途径。因此本文对两种思维方式之间最重要的、最不可调和的差异做了总结。然而，由于本文是在科学可视化研讨会上提出的，在此关注点将集中在信息可视化的理念与技术。
2. Principal Differences （主要差异）
Card et al. [1] defines the two forms of visualization as:
Card等人[1]将两种形式的可视化定义为：
- Scientific Visualization: the use of interactive visual representations of scientific data, typically physically based, to amplify cognition.
- Information Visualization: the use of interactive visual representations of abstract, non-physically based data to amplify cognition.
- 科学可视化：（通常是基于物理的）科学数据的交互式视觉呈现以加强认知
- 信息可视化：抽象的，基于非物理数据的交互式视觉呈现以加强认知
While Scientific Visualization covers accurate visualizations of the real world, Information Visualization covers visualization of concepts that often are abstract in nature. The purpose of Information Visualization is to make it possible for analysts of data to obtain internal mental models of the information content in datasets;models which subsequently can be used for characterization, prediction, and/or decision making.
科学可视化涵盖了现实世界的精准可视化，而信息可视化覆盖了自然界中通常较为抽象的概念。信息可视化技术的目的在于数据分析来获取数据集信息内容的内在思维模式变为可能，这些模式随后可以用于定性，预测，和/或者制定决策。
Keim et al. [2] classifies visualizations into the categories:
Keim等人[2]将可视化进行了下面的划分：
- Exploratory Analysis: the search for hypotheses
- Confirmatory Analysis: the confirmation or rejection of hypotheses
- Presentation: the presentation of facts that are fixed a priori
- 探索性分析：对于假说的探寻
- 验证性分析：对于假说的证明或者推翻
- 展现：先验事实的展现
2.1 Information Visualization and Exploratory Analysis （信息可视化和探索性分析）
This example is completely imaginary.
下面这个例子是完全虚构的。
A grocery store sells Feta-cheese for use in salads. It notices that it is losing money on this product and therefore decides to stop selling it. However, after having stopped selling this product, the grocery store notices that it now loses far more money than it did before. The reason for this is that the few customers who bought this product were wealthy and also bought many other products at the grocery store. When the store stopped selling the product, the customers went to other stores where they could buy it and, at the same time, purchased all other products they needed there.
一家杂货店销售沙拉用的羊奶酪。由于发现产品在赔钱因此决定停止销售。然而，当停止销售此商品后，杂货店发现赔的钱比原来更多了。原因是购买羊奶酪的顾客中的一些人比较有钱并且会同时在杂货店中买许多其他的东西。当商店停止销售改产品时，这些顾客去其他商店买羊奶酪，同时在那里买其他的商品。
The grocery store has modern cash-registers and records all purchases in a large database, so the information about the customers’ purchase patterns were, in principle, available before the store made the erroneous decision of stopping to sell Feta-cheese. An Information Visualization analyst studying the grocery store’s database, in the state that it was before the erroneous decision was made, should by performing an Exploratory Analysis, in principle, have been able to foresee this, without having been told anything about the data in advance.
这家杂货店在拥有现代化的现金出纳设备，并且将所有的支付记录保存在一个大型数据库中，因此顾客的支付模式信息，在原则上，在商店做出停止销售羊奶酪的错误决策之前就已经可用。如果有一名信息可视化分析师研究了杂货店的数据库，在错误决策做出之前通过执行探索性分析，即可在预先不知道这些后续数据的情况下预见到杂货店的损失。
2.2 Scientific Visualization and Confirmatory Analysis （科学可视化和验证性分析）
This example is from the IEEE Visualization 2006 conference, where a visualization contest will take place. The subject of the contest is the TeraShake 2.1 earthquake simulation data set and the task is, through visualizations, to answer five questions, such as ”Which regions produce wave reflections?”. There is thus, in this case, a clear hypothesis that some of the regions will produce wave reflections and this hypothesis must either be rejected or confirmed by finding these regions. Using the terminology of Keim, the task is therefore to perform a “Confirmatory Analysis” of the data.
这个例子来源于IEEE可视化会议2006，会议举行了可视化竞赛。竞赛的主题是TeraShake 2.1地震模拟数据集，任务是通过可视化回答5个问题，例如“哪个区域产生了波反射？”因而在此例中，一个清楚的假设是一些区域会产生波反射，并且通过探寻这些区域该假设或证实或推翻。使用Keim的术语，该任务实在执行数据的“验证性分析”。
3 Common Visual Data Exploration Techniques （常见的视觉数据探索技术）
In the following is listed a few examples of the techniques that are used in Information Visualization to explore datasets that initially are completely unknown.
下面列举了一些信息可视化中探索初始完全未知数据集使用的技术
3.1 Icons and Glyphs (图标和字形)
Markers are featureless objects for representing positions in scatter plots, but multiple, clearly distinguishable markers are sometimes used in one and the same scatter plot to visualize important information. While markers usually only have positional attributes, there are objects that are designed to represent many pieces of information simultaneously. Such objects are called icons or glyphs ([3,4]) and have attributes, such as position, size, shape, color, and orientation, to which data is mapped. The hope is that users, by visual exploration, may be able to find patterns among the objects.
标记是表示散点图上位置的无特征对象，但是多重的，可以清楚区分的标记有时在同一张散点图上对重要的信息进行可视化。标记同行只有位置属性，而有一些对象被设计用来同时表示许多条信息。这样的对象成为图标或者字形 ([3,4])，并且拥有诸如位置、尺寸、形状、颜色和方向等属性，数据被映射在这些属性中。目的在于用户通过可视化探索，也许可以从这些对象中发现模式。
3.2 Brushing & Linking （关联更新技术）
Brushing & Linking is a widely-used visual data exploration technique, with which multiple different visualizations of a dataset are viewed simultaneously. This technique is often used to combine the advantages of multiple forms of data visualization techniques, but can also be used to e.g. allow simultaneously brushing of markers in multiple scatter plots, organized in a scatter plot matrix. The advantage is that the interaction involved is both unambiguous and simple to perform and therefore seem familiar and natural to most human beings. In the beginning of a visual data exploration process, all markers have a common color. The process consist of
关联更新技术是一种广泛使用的视觉数据探索技术，通过它多种不同的数据集可视化能够同时观察。此项技术经常被用于将多种形式的数据可视化技术组合起来，但也可被用于，比如说，同时更新组织在散点矩阵中的多个散点图中的标记。其优势在于参与的交互明确而易于执行，因而使大多数人来说感到亲切而自然。在视觉数据探索步骤的开始，所有的标记颜色都相同。过程包括：
- Selecting one of multiple views.
- Brushing selected markers in this view with a color.
- Inspecting the other (linked) views to see the effect
- 从多个视图中选一个
- 将视图中的标记着色
- 检查其他视图看看效果
3.3 The Grand Tour (总览)
In “The Grand Tour” technique [5–7] high-dimensional data are considered as forming a shape in a high-dimensional space. and the method therefore consists of moving a 2D or 3D “window” around this shape, so that a low-dimensional projection of it becomes visible. Significant features of the high-dimensional shape may thereby become distinguishable, and, if one is lucky, one may even be able to guess what kind of high-dimensional shape the data form.
在总览技术中 [5–7]高维度数据被视为存在于高维空间中的图形，因而此方法包含在形状中移动一个2维或者3维的窗口，以至于其低维度投影变得视觉可见。高维度形状的有意义的特征因此可以观测得到，并且如果足够幸运，甚至还能够以此猜测数据形成的是什么样的高维度形状。
This technique is often combined with “Brushing & Linking”, so that subsets of the markers at any time may be brushed with a color to make them distinguishable from the other markers.
此技术通常与“关联更新技术”组合起来，以至于标记的子集在任何时候都可以被更新为易于和其他标记分辨的颜色。
4 Software （软件）
There are also differences in the software that typically is used for Scientific Visualization and Information Visualization. Consider for example the Visualization Toolkit (VTK) [8], which is an open source visualization system that dates back to 1993. It was designed to be robust, understandable, extensible, modular and maintainable. For that purpose, it offers scientists a consistent object-oriented programming model that in 1993 was considered sophisticated, where visualizations are created by linking together a series of objects into a pipeline.
通常被用于科学可视化与信息可视化的软件之间也存在着不同之处。例如考虑可视化工具包（VTK），它是一个能够追溯到1993年的开源可视化系统。它被设计为健壮的，可理解的，可拓展的，模块化的和可维护的。为此目的，它于1993年为科学家提供了一个在当时被认为复杂的一致的面向对象编程模型，在此模型中可视化由一系列对象一起链接在管道中创建。
Seen from an Information Visualization perspective, it is clear problem that the visualization pipeline is designed to be “single-pass”, where only the resulting visualizations can be interacted with and where interaction usually is limited to rotating the visualization. Information Visualization techniques, such as Brushing and Linking require continuous interaction with the a database containing the original dataset. This requirement is not listed above, which makes VTK unsuitable for Information Visualization.
从一个信息可视化观点看来，可视化管道设计为“单通道”是一个明显的问题，基于该设计只有产生的可是效果才可交互，并且交互通常被限制在可视化图形的旋转上。信息可视化技术，例如关联更新技术，需要与包含原始数据集的数据库进行持续交互。这个需求没有在上面列出，这使得VTK不适合于应用于信息可视化。
However,many Scientific Visualization analysts study “ Computational Steering,” with which feedback can be given to running simulations, based on visualizations of early results from the simulations. The requirements for these kinds of software therefore approach the requirements that Information Visualization scientists place on the software they use.
然而，许多科学可视化分析师在研究“驾驭式计算”，通过它可以给运行模拟器反馈，基于模拟器早期结果的可视化。因而这些类型软件的需求接近了信息可视化科学家们对于他们所使用软件的需求。
5 Conclusion （结论）
This paper presented Scientific Visualization and Information Visualization with emphasis on the latter. It describe a number of Information Visualization visual data exploration techniques and finally compared the requirements for the software used in both disciplines.
本文阐述了科学可视化与信息可视化并着重阐述了后者。论文描述了一些信息可视化视觉数据探索技术并在最后比较了这两个学科使用软件的需求。
References （引用文献）
1. Card, S. K., Mackinlay, J. D., Shneiderman, B.: Readings in Information Visualiza- tion: Using Vision to Think Morgan Kaufmann (1999)
2. Keim, D. A., Ankerst, M.: Visual Data Mining and Exploration of Large Databases Tutorial, Conf. on Principles of Knowlege Discovery and Data Mining Freiburg, Germany (2001)
3. Carr, D. B., Nicholson, W. L.: Evaluation of Graphical Techniques for Data in dimensions 3 to 5: Scatterplot Matrix, Glyph, and Stereo Examples Proceedings of the Section on Statistical Computing (1985) 229–235
4. Pickett, R. M., Grinstein, G.: Iconographic Displays for Visualizing Multidimen- sional Data Proceedings of the IEEE Conference on Systems, (1988) 514–519
5. Asimov, D.: The grand tour: A tool for viewing multidimensional data SIAM. Jour- nal of Science and Statistical Computing 6 (1985) 128–143
6. Buja, A. and Asimov, D.: Grand tour methods: an outline. Computing Science and Statistics: Proceedings of the Seventeenth Symposium on the Interface (1985) 63–67
7. Wegman, E. J.: The grand tour in k-dimensions Computing Science and Statistics: Proceedings of the 22nd Symposium on the Interface (1991), 127–136
8. Schroeder, W.J., Martin, K., Loresen, W.: The visualization Toolkit: An Object Oriented Approach to Computer Graphics, Third Edition Kitware, Inc. (2004)
本文链接：http://bookshadow.com/weblog/2014/05/11/scientific-visualization-vs-information-visualization/
请尊重作者的劳动成果，转载请注明出处！书影博客保留对文章的所有权利。
捐赠书影博客
