The first half of the paper is concerned with giving background about information visualization and defining a basic vocabulary for the field. The second half of the paper discusses the process of mapping data to visual form.
The paper begins by asserting that visual external aids help humans achieve expanded intelligence by assisting them in solving logical problems. While graphical aids for thinking have been in use for a very long time, computers have taken them to a new level. Through their use, huge amounts of data can be analyzed dynamically and interactively at a low cost. The authors define information visualization as the use of computers as graphic aids in the areas of business, scholarship, and education.
The first section attempts to explain the concept of external cognition through examples. The authors define external cognition as the use of the external world to achieve cognition. The first example they present in the multiplication aids subsection is longhand multiplication. I thought this was a great first example since it shows a situation in which external aids make it clearly easier to accomplish a task, or as the authors put it, they amplify cognitive performance. In this case, the visual aid is used to expand a person's working memory. The authors also mention the slide rule as another visual representation for multiplication-although it provides the answer through interactive manipulation.
The subsection on multiplication aids then moves to a confusing example of a nomograph used to calculate a slightly complicated physics problem. The point of the example is to show how visual aids can be used as computational devices beyond simple multiplication. The last example in the subsection is the Graphing Calculator, a computer application that graphs functions in three dimensions while performing possibly millions of mathematical operations. The graphing allows the user to gain a further understanding of the properties of the function that could hardly be achieved by looking at numbers.
A short subsection on navigation charts goes over different kinds of projections of the Earth that are used in nautical charts, each with its own advantages and shortcomings. These charts are examples of visual aids used both as calculators and storage devices because they help calculate location and heading of a ship while also providing the shape of the coastline, political boundaries, etc.
The subsection on diagrams begins by asserting that some diagrams are not particularly useful. The authors use the painful example of a poorly designed diagram that was used to decide the launch of the space shuttle Challenger in its last and fatal mission. The problem with the diagram was that it didn't give spatial priority to the two most important variables in question. An alternate diagram that uses the two most important variables in the x and y-axis clearly shows a pattern that would have told NASA not to launch the Challenger. The lesson is that the two most important variables should always go in the x and y-axis.
The next diagram shows information on sleep and feeding habits of babies in their first 6 months of life. The clever setup of the diagram shows a clear pattern developing after about 17 weeks (I guess that's when the parents can start sleeping normally again). Another diagram also shows patterns by displaying information about tides.
The authors end the external cognition section by asserting that the invention of visual artifacts has been crucial to the progress of civilization. The punch line is that the next generation of visual artifacts will come from computers (i.e. it will come from visualization).
The section on information visualization attempts to define what information visualization means and gives some examples. Before defining information visualization many terms are defined. There's a handy table that lists the definitions at the bottom of page 7. Information visualization is defined as the use of computer-supported , interactive, visual representations of abstract data to amplify cognition. The introduction to the section also mentions some of the challenges information visualization faces mainly because of the abstract nature of the data it deals with. The first subsection provides some background on the history of the field. The next three subsections provide examples of information visualization. The first is an interactive periodic table. The second example is an application that handles real time financial data used for risk-hedging. It is able to show in one screen what previously took 100 screens. The third example is an application used to detect telephone fraud that takes thousands of data points and makes sense out of them by plotting them in a clever manner. The point of the examples is to show the added power diagrams get from being dynamic and/or interactive.
The section on cognitive amplification starts by explaining the concept of knowledge crystallization. It gives an example of someone having to do research on laptops available in the market. It then lists the six steps involved in knowledge crystallization on page 11 (they can also be found in figure 1.15 on page 10). The goal of a knowledge crystallization process is to get the most compact description possible for a set of data relative to some task. The authors affirm that information visualization can help most parts of knowledge crystallization. The subsection on visualization levels of use defines four levels at which visualization can be used. These are summarized in table 1.2 on page 14. The next subsection defines cost of knowledge-the amount of knowledge you can get given a certain amount of time. The authors assert that some visualization tools are better for dealing with small amounts of data and others are better at dealing with bigger amounts of data. Table 1.3 on page 16 provides a summary of the last subsection by listing six major ways in which visualization amplifies cognition-including examples.
The section on mapping data to visual form deals with database theory. The subsection on data tables seems to be right out of a database book. It relates the process of getting raw data in tables, and then running queries (the authors call them data transformations) on them to get information that is more relevant. It even goes retro by mentioning hierarchical and network data models-these are relics from the 60s and 70s. The subsection on variable types says that data comes in three different types: nominal, ordinal, and quantitative. The next subsection defines metadata as data about data. The subsection on data transformation defines four different types of data transformations (they look like queries to me) and gives some examples. This section oversimplifies issues in database theory, and doesn't pay much attention to the relational model.
The section on visual structures goes into the biology of the eye and recommends effective techniques for displaying data. This is the section I enjoyed the most because I learned the most from it. It gets to critical matters in the field of information visualization. The introduction states that data tables are mapped to visual structures. Visual structures are expressive if they show all and only the data in their respective data table, and effective if they are faster to interpret.
The subsection on perception gives some background on how the human eye works. It provides a good deal of unnecessary numerical information. An important point in the subsection is that visual structures should be designed so that the eye can use automatic processing when looking at them. Automatic processing is the fastest mode in which the eye can process information. Some patterns the eye will distinguish are listed in table 1.20 on page 26.
The subsection on spatial substrate tells us that there aren't many ways of mapping data tables to visual structures since visual structures are made from spatial substrate, marks, and the marks' graphical properties. It points out that the use of space is paramount because space is perceptually dominant. Since space is so important, the authors then describe five different techniques that can be used to increase the information encoded in space.
The subsection on marks defines the four elementary types of marks: points, lines, areas, and volumes. The following subsection explains how points and lines can be used to form trees and graphs. The subsection on retinal properties covers other ways of adding information to a visual structure. For example, marks can have color, size, texture, and shape. All three tables on page 30 are useful in showing all the different options. Table 1.23 is particularly useful because it shows the effectiveness of properties for different types of data. The problem with this table is that you have to use controlled processing (i.e. read the description above the table) in order to know whether the empty or filled circles are good. The subsection on temporal encoding says that humans are very sensitive to changes in mark position. It then goes on to talk about the representation of data that change with time and animation.
The section on view transformations defines them as ways to modify and augment visual structures and goes on to explain three common view transformations. Location probes use location in a visual structure to provide extra information. Viewpoint controls allow users to navigate through a visual structure. This subsection has a strong Maryland flavor with its two examples being zoomable interfaces and the overview + detail approach. Distortion can also be used to create focus + context views.
The section on interaction and transformation controls states that humans can interact at any point of the visualization process from raw data all the way to view transformations as shown in figure 1.23 on page 17 and gives some examples.
This paper should enable scholars interested in the subject to share a common vocabulary when communicating. It provides the foundation for a healthy development of the field. It also gives a summary of background information and basic techniques used in the field that can be useful to people with interest but little knowledge in information visualization.