One of the most important ways to represent knowledge is the document written in natural language. Communicating information through language is represented along a one dimensional axis, even if represented in a two dimensional document. Therefore, the time required for communication is approximately proportional to the number of words. In the past, when language communication was the only means, serial communication was reasonable. Serial communication allows us to perceive one phrase followed by another, which is appropriate for intensive reading, like novels, self-paced training manuals or technical literature.
Information in newspapers or business documents, however, often covers wider topics selected by the reader. These documents differ greatly from those listed above, which need to be easy to read and understand. Here, the important information must be apparent at a glance or by skimming. Modern documents often combine both types.
Even within one document, when there is a lot of information, we first want to summarize and to skim the document. We like a powerful browser that covers multiple pages at one time. Tools like a table of contents, keyword index, online retrieval, and global document layout viewer such as a newspaper are all quite effective.
The trained human, as a receiver of language information, can process multiple information sources simultaneously. We can train ourselves to understand multiple conversations at the same time. Speed reading is also well known, with which we scan the page in two dimensions.
In general, a larger page makes it easier to grasp the contents. We are able to grasp the main topics of a newspaper page at a glance. I believe most people like to use a larger map than a small one. We will soon be able to use large screens like those in Figure 1. How can we better use this environment to represent documents?
As I mentioned above, a written document converted from aural language allows quicker understanding of the content, because of high speed visual perception as well as quick global grasp by quickly scanning a two dimensional screen. Shortening character lines, combined with multiple column layout, is also effective for speed reading. See Figure 2(a), 2(b) as examples. A sophisticated automatic column layout is not provided by software on the market yet. People can do it, however. The most optimized text layout is now used in newspaper layouts put together by professional designers.
Fig.1: When large screens are available in the near future, what kind of text layout will be appropriate?
Here are small experiments. First, we show an example in Figure 2(a) as solid typesetting. These two pages of solid typesetting are modified into three column typesetting with smaller fonts, as shown in Figure 2(b). Now, since all sentences and images are on one page, we can view the whole content at a glance. This procedure is achieved automatically.
In Figure 3(a), the layout unit has been changed from chapter to paragraph. With this small modification, view-ability has improved. Using smaller units of the text as shown in Figure 3(b), the characteristics become clearer. Even if some parts of the paragraphs are hidden, the general story is still understandable.
If you want to experience the differences in readability and the speed of grasping whole content among these types of layouts, print out the attached PDF files and compare them with each other.
Taking this strategy further, text boxes are arranged first by screen size as shown in Figure 4. Here, the uppermost layer A has only three text boxes, but the second layer B shares a larger area and has ten boxes. The lowest and largest layer has 54 boxes. The layer structure and number of boxes will be decided depending on the volume of text, number of paragraphs, font size and screen size.
Fig. 4: Multi-layer text boxes
Text layout has a long history. Figure 5 is an ancient example used in the Maya Dynasty. Mayan characters are mainly used for inscription articles to commemorate a ruler's record. Considering the ancient social background, in which very few people can read, characters are delineated one at a time by pointing a finger. It may be reasonable for an adjacent character to be placed in a neighboring area. As a result, Mayans employed zigzag reading order as two character columns combined together (M. Coe, M. Stone, "Reading the Maya Glyphs," Thames & Hudson, Nov., 2001).
Fig. 5: Mayan characters
Until a few years ago, people thought that high-resolution displays could not be made larger while bringing down costs. In reality, however, we see that the both are possible. And we now use not only text but also diverse digital contents, which are linked with each other. Representation of knowledge is a most interesting research topic, which will be accelerated.