Implementing an Insight Center:
Hypertext Navigating
Hypertext is defined in different ways by different people, but there are a number of characteristics common to most hypertext systems and these can be used to provide a definition of hypertext. Hypertext is, generally speaking, a way of overcoming the normally forced sequential nature of printed textual material. In most printed documents, one reads the first sentence, then the second, then the third, etc., in order until the document is finished. In contrast, in hypertext, the reader/viewer has the power to decide what to read next, the level of detail desired and where to move next in the text. Hypertext is then nonsequential; there is no predetermined order to a "text". Text is thought of as a unit of meaning - such as a piece of actual text, text and pictures, a report, a book, a collection of books, etc. It could include almost anything representable on a computer - such as data, sound, and video. It can be a compound document or multi-media presentation of some sort. Hypertext is sometimes called "generalized footnotes" since the reader of a normal book decides if and when to read the footnotes in the book.
Hypertext is, at the fundamental level, a body of interwoven pieces of information. The pieces of information are called nodes and contain pointers, called links, to other nodes. A link points from one node of information, called the anchor node, to another called the destination node. Figure 1 contains an overview of a possible hypertext with various links between various pieces of text. The user moves from one node to another via the links. The presentation of nodes and links varies from system to system. In some, a node is shown as a screen full of information, with various ways of representing "hot" spots, the links, that change the screen of information when selected. In others, the nodes are represented as windows on the screen and the links are again some "hot" spot or button representation that, when clicked, pops up more windows on the screen. In many systems there is a backtrack facility that allows a user to back up to a previously visited node.
Figure 1
One of the most important aspects of hypertext is the one that is hardest to define:
Many non-hypertext computer techniques may at least match various aspects of the definition of hypertext, but true hypertext should also make users feel that they can move freely through the information, according to their own needs. This feeling is hard to define precisely but certainly implies small overhead with respect to using the computer. This means short response times so that the text is on the screen as soon as the user asks for it. Small overhead also requires low cognitive load when navigating, so that users do not have to spend their time wondering what the computer will do or how to get it to do what they want.
This feel of movement and flexibility brings on descriptions of it such as navigating an information space, or browsing a hyperspace, etc. These systems provide a way of interacting with them that makes use of something humans do well: absorbing information laid out along two-dimensions, i.e. diagrams, pictures, diagramic metaphors, etc. There is a transparency of use associated with such systems. It is relatively easy for first-time users to understand how to use the system via trial and error, a process of discovery. One just gets in and drives. The amount of required specialized a priori knowledge, such as needed for formulating queries for a database system, is kept small. One can start with a sketchy, limited mental model of the system and, through use, fill in the gaps in that model.
The Analyst is a hypertext information management system, a paperless office system. It is capable of maintaining, linking and retrieving information in many forms such as graphics images, spread sheets, and finished reports. The system makes use of a "container" model that is based on a generalization of the office file cabinet, called an Information Center in the system. The Information Center contains folders, and each folder can, in turn, contain additional folders:
The Analyst organizes data in a method designed to be both intuitive and flexible. The system is easy and natural to use for anyone familiar with the operation of the standard office filing system, as its design and terminology are based on those very concepts. Just as the usual filing scheme in a office consists of a hierarchy of filing cabinets, drawers, and folders, the Analyst organizes large collections of data into Information Centers that are further organized into smaller collections called folders, which in turn contain items and other folders.
This container metaphor is similar to the directory-file system of UNIX and DOS systems; it is a hierarchical, tree-structured arrangement of information. The power and flexibility of the Analyst comes from the abstraction of this metaphor. The fact that a folder contains some item or another folder is, in lower level computer terms, merely a pointer from the first folder to the other item or folder that is interpreted or labeled "contains". The Analyst allows the user to create and manipulate labeled links between items in the system. The links are bi-directional and provide a way of navigating through the Information Center to find desired information items. The system makes use of an explicitly represented semantic network that the user constructs. This construction is performed through either a graphical node-link editor, or with a tool called the Assistant, that allows the creation, manipulation and deletion of links according to certain rules.
As said above, particular items of information are located in the system by navigating across labeled links. These are presented to the user in the form of pop-up menus with the link labels and their destinations as menu items. As the user moves the mouse pointer to the proper label and then to the right, and as the pointer crosses the menu's right border, the system pops up a menu of all links that point to that particular item or have that item as an anchor, while leaving the previous menu up. This process can be continued by moving to the right to pop up more menus of links. The system centers the whole collection of displayed menus as the user reaches the right-hand screen boundary. The process can be halted by releasing the mouse button on a menu item. The system then pops up a window containing that item or folder, and the menus with the links disappear.
Figure 2 is a screen capture from the Analyst showing a demo Information Center that comes with the system. The partly obscured window in the upper left is the Information Center, labeled "ADEMO Info Center". The contents of the Information Center are listed in this window, such as "Graphics". The icon to the left of the word "Graphics" in this window means that it is a folder. If we "open" this folder by selecting it we get another window, the one laying directly on top of the "ADEMO Info Center" and labeled "ADEMO Graphics". Its contents include three folders including the folder "Images." If that folder is opened, the window labeled "ADEMO Images" appears. It contains a list of the contents, which includes the graphic (represented by the camera icon) "Pepper." When this is opened, a graphic editor containing the image of a pepper appears. (The pepper window is partially off the screen in order to show the other folders and their contents more clearly).
Figure 2
Figure 3 is the same as Figure 2, except that it shows the navigating menus. The user can go directly to the pepper image without having to open all the folders in between. Notice that the links are "contains" links (shown in parentheses on the menus). The use of the bi-directional nature of links can be seen on the bottom of the middle menu. The word "contains" is to right of the item listed, in this case, "ADEMO", the Information Center itself. This indicates that there is a "contains" link that originates from "ADEMO Info Center" to the folder labeled "ADEMO Graphics". If this link were selected it would return the user to the "ADEMO Info Center".
Figure 3
This system of menus is generated automatically from the underlying semantic network that specifies folder contents via "contains" links. That is, once the Analyst is given a semantic network, it generates a series of cascading menus based upon the "contains" links in the semantic network.
Figure 4 shows a graphical way of interacting with Analyst. The part of the underlying semantic network of the Information Center "ADEMO" is displayed in a window in the foreground. This shows only the "contains" link from the folder "Graphics" to other Information Center items. By clicking on a node, the user can open up that node. The user can scroll the node-link graph and bring to view the areas that lie outside the window. The user can also filter out what links s/he would like to see. This helps to manage information overload by allowing the user to see only the desired links and nodes. The nodes and links can be manipulated, created and deleted from this window.
Figure 4
The Analyst comes with a rich set of data and information representations. There are spreadsheets, databases, chart editors, graphic editors, and maps. There are simple text items, outlines and documents. The document system makes use of full text search and keywords. The keywords can be linked into the underlying semantic network. Documents can contain things called forms that allow the insertions of any sort of graphics, text, Smalltalk code, links to the semantic network, buttons that activate various system functions and even animation sequences. So it is possible to build very complicated interactive pieces that are highly linked to anything else in an Information Center. Everything is renderable into Post Script and therefore hard copy for anything in the Analyst is available. There is also an expert system tool called Humble that can be added to the system to develop knowledge-bases. Finally, since the Analyst was built on top of the Smalltalk-80 system, a knowledgeable Smalltalk developer can add to the system any feature desired.
One aspect of the first version of the Analyst is limiting. The developers of the system apparently made no allowance for the automatic entering of information. The user of the system must enter everything by hand. In particular, there is no way to automatically generate the underlying semantic network from the data in the database. This forces the user of the system to enter the semantic network "by hand" or to augment the Analyst system with additional programs written in Smalltalk. This limitation was overcome by going through the Analyst source code and extending the system enough to generate automatically the underlying semantic network in an Information Center from a database table.
In summary, the Analyst offers the following:
