What is orientation?
A hypermedia structure is a model of some domain, existing or imaginary. This model is usually represented in the system as a network of nodes and links. The user can search for information by moving around in this structure.
By orientation we will mean 'knowing where you are' in a hypermedia structure, in a spatial sense, like the third page of the book, the second street after Main Street, or the middle of the film. Even though the last example involves time, it can be thought of as a position in some kind of space. This 'space' does not have to resemble a physical space. It might, for instance, be a space which is made up of a structure of concepts, like Ancient Roman Law or the history of the Greek Gods.
A user browsing through a hypermedia system might know for certain that she is reading about some aspect of Victorian Art, but she might not know how this concept is related to other concepts and sub-concepts. Thus, she might know where she is and what she is presently reading, but not how that is related to other concepts and subjects. In this case she is not oriented. In order to be oriented we must know where we are in relation to something. This 'something' can be thought of as some kind of whole which defines a frame of reference, a structure of information items in which orientation takes place.
In order to orient herself the user must have certain knowledge of the world and its reference system. She must have a cognitive map of the world. Dillon et alia (1990) describes three stages in how a cognitive map is acquired:
Survey knowledge is essential to orientation. An overall view of the hypermedia system is needed to minimize the risk of getting lost. Here the term world vision will be used in a meaning similar to survey knowledge (see discussion below). Having a general overall knowledge can be compared to the concept of pre-understanding in the reading process. If the reader has a general idea of what the text is about reading is much more fluent. With a high pre-understanding, there is a high probability that the reader makes correct guesses while reading and thus do not have to stop and think about every word in the text (Edfeldt 1982).
The user needs to have an overall view of the structure of the subject area, the structure of the hypermedia network, and the structure of the functions for moving around the network. Furthermore, successful orientation requires that the user knows the current location in the hypermedia structure and how this part is related to the whole. We will call the information necessary for orientation for orientational information. Disorientation occurs when the user lacks orientational information.
The anatomy of a hypermedia system
We have stated that orientation takes place in relation to some kind of world or system, the hypermedia system as a whole. The different parts of a hypermedia system can be described in the following way:
As an example of these different elements, consider the process of authoring a book. The author might start with doing research on the subject she is going to write about. It is possible that the subject area is already structured in some generally accepted way. The author then decides how to communicate that information to the reader, that is how to present the subject in the form of a book. The way the author struc-tures the presentation in the book might differ from the structure of the subject. For instance, there can be complex relations which is not possible to present straightforward, but require lengthy discussions and many chapters to explain. There might also be pedagogical reasons for introducing concepts in a particular order. In a hypermedia system we have a similar situation, we have a subject area which we want to communicate to the user.
Disorientation problems can arise in the context of these different areas: the subject (that is the contents) and the presentation of the subject (that is the form). The user might get lost because she does not understand the structure of the subject area. She might also get lost if she does not understand how the hyper-media network is structured. Thus, we can differ between conceptual and spatial disorientation (Gygi 1990).
Furthermore, the user might get lost if she does not know how to operate the user interface for accessing different parts of the structure, that is the browsing and navigation functions. This last problem is less common in traditional media. An interactive system is special in the respect that it is more complex to operate than, for instance, a printed book. The only browsing/navigation functions of a book is to turn pages, one or many at a time, forwards or backwards. A hyper-media system has the potential for a much richer set of possible ways to access the material.
Orientational information can be viewed as the part of the user interface which helps the user to find her way around the database. A book also has orientational information, like the table of contents, index, page numbers, foot notes, et cetera, which must be known by the reader in order to access the material in an easy and efficient way.
It is important how the user interface is designed, that is how the parts of the interface are presented to the user. In general, information can be presented on different levels of abstraction: the physical level, the pictorial level, and the symbolic level. A written text, for instance, is a symbolic representation of an object or a concept. The word "Dandelion" is an abstract symbol for the Dandelion. A picture is more concrete, and the most concrete level is the Dandelion itself.
Which kind of representation is communicated to the user will affect her under-standing. In this thesis it is argued that explicit visual presentation forms are desirable for communicating orientational information. This is in part related to our view on browsing and navigation - 'physical' movement in a hypermedia structure - which implies some kind of spatial representation of the network. If the subject area is about concrete objects it makes sense to show orientational information pictorially, like a map. This is also related to the use of metaphors in user interface design (Dillon et alia 1990). Navigational metaphors, like maps, guided tours et cetera, can help building a cognitive model of the system.
However, there are abstract subjects which might not be straightforward to visualize. Concrete representations can therefore be inappropriate. Moreover, a picture can be ambiguous in some cases. If the user points and clicks at a photo of a Dandelion, what information is she asking for? The specific Dandelion she pointed at? Dandelions in general? The Dandelions at the place the picture was taken?
World vision versus tunnel vision
An overall view of the world, in our case a hypermedia system, and the major points of reference in this world are vitally important for orientation. This is also known as world vision (Casey 1986). The opposite to world vision is tunnel vision. Tunnel vision, that is lack of a high level picture of the system's struc-ture, is one basic reason for disorientation problems (Casey 1986). Tunnel vision is caused by, among other things, information being presented in small fragments, out of context.
The context is the setting in which an item of information appears. The context helps the user to remember where she is and provides a frame for the interpre-tation of information. Reference points is one example of contextual informa-tion. There are many different kinds of context. Context can be related to con-tent, form, space, time et cetera (Petterson 1989). Here focus will be on the inner context of a presentation, such as the layout on a page and the interplay between text and pictures. An example is comic magazines, where the text is presented in its context using voice balloons.
In a hypermedia system nodes represent items of information and links repre-sent relations among those items. There is a risk that the node-link model can result in a fragmented conception of the systems' structure, by tempting, or even forcing, the author to divide the material into small elements. This can be espe-cially true if the hypermedia tool used to implement the system places restric-tions on the size of a node, and/or limits the presentation to one node at a time only.
In addition, restrictions on available screen space often makes it difficult to fit information on one screen, forcing a separation of related information across different screens. Having to switch back and forth between the nodes can cause disorientation, since it is hard to remember which context a node belongs to if no orientational information is provided. Figures 3.1 to 3.3 illustrate this problem.
Figure 3.1: An overall view of the solar system. Clicking at the planet Venus (the second inner planet from the Sun) causes the screen shown in figure 3.2 to be displayed.
Figure 3.2: Presenting information out of its context can cause tunnel vision. Without the reference points provided by the map of the solar system in the screen in figure 3.1, it is difficult to relate the detailed picture of Venus to the whole.
Figure 3.3: World vision - details related to the whole. A solution to the problem in figure 3.2 could have been to present a miniaturized version of the map together with the detailed picture.
If no contextual information is given, the user has to remember the overall structure and how different nodes are related. Since short-term memory is limited (Waern 1989), the risk of getting lost is obvious. Rapid browsing can increases this risk further, since it becomes even harder to remember the structure traversed.
Even if it is possible to simultaneously view multiple nodes displayed in separate windows, as in, for instance, NoteCards (Halasz et alia 1987), problems may arise. Confusion and disorientation can result if the nodes are not visually related, and if no contextual information is given.
The problem of fragmentation is somewhat similar to the problem of modes in interactive systems (Krona 1990), where the user is restricted to a particular part or function of the system at a given time. Modes are notorious for causing disorientation and confusion (Smith et alia 1982). While windows have partly solved the mode problem (Tesler 1981), hypermedia systems displaying only one node at a time have reintroduced some of the original problems involving modes.
Complex structures can disorient the user
The author of a hypermedia system has the responsibility for structuring the database in a clear way, so that the user can understand it. Several aspects must be considered: the purpose of the system, the intended user category, typical tasks, style of browsing, et cetera.
Since nodes and links can form a network structure of arbitrary complexity and size, it is possible to create a hypermedia system which could be very confusing and difficult to navigate, no matter how well orientation is aided. Besides being intricate and mysteriously organized, the network might be structured differently from the structure of the subject area. Clearly, this can cause disorientation.
Disorientation problems could be reduced by decreasing the structural complexity, for instance, by using sequence and hierarchy only, and avoiding complex webs. This has been studied in previous research where authors have reported that in some cases a hierarchical structure is preferable to a mixed structure (Edward & Hardman 1989).
An additional way to address problems caused by intricate networks and by having a confusingly high number of paths to choose among, is to use predefined guided tours through the material (Marshall & Irish 1989). Another approach is to utilize path mechanisms which keep track of in which order information has been accessed, thereby making it clearer which nodes have been traversed (Nielsen 1990). Bookkeeping thus compensates for lack of visual overview. In chapter 5 such techniques will be discussed in greater detail.
The abstract nature of hypermedia
There are several properties of hypermedia which make it potentially more dif-ficult to comprehend than non-computer based media. Hypermedia and printed media have been compared previously by, among others, Yankelovich & Meyrowitz (1985), and Broady (1990).
Information stored in a computer can be more abstract and more difficult to grasp than, for instance, printed informa-tion. When we pick up a book or a magazine, we can instantly perceive how thick it is, see how much text it contains, if there are any illustrations, et cetera. Just by flipping through the pages of a book we can get a lot of information about its structure and contents. Furthermore, tables of con-tents, page numbers, headings, subheadings, and other typographical cues, help the reader to orient herself (Evenson & Rheinfrank 1989). Similar principles are not yet systematically used in hypermedia systems (Broady 1990).
When we encounter a particular hypermedia system for the first time we might have difficulties estimating the, potentially large, amount of informa-tion enclosed in the system, and we are likely to have only a vague idea about how it is structured. It might be difficult to tell how much information is con-tained in a hypermedia system since we can not perceive its volume physically. Even a small system can be difficult to comprehend. Thus, while large storage capacity is a major advantage of computerized media, it can also give rise to new problems and difficulties.
Furthermore, the high degree of freedom and interactivity inherent in hypermedia systems can cause disorientation (Marshall & Iris 1989). Even though a book can be written in an unclear and fuzzy style, the physical location of the pages does not alter. Information in computer based systems on the other hand, can be organized and accessed in a variety of ways. Hypermedia makes it possible for the user to choose among numerous paths, increasing the risk of getting dis-oriented. Indeed, one of the basic ideas of hypermedia is that the user should be able to access information in multiple ways.
However, ease of moving around can also help in building a conceptual model of how the system is structured. Being able to scan, explore and browse rapidly in different directions to see where different trails end up could help in getting an overview (Larnhed 1989). Compare this to, for instance, quickly browsing through a book, or to fly across a landscape in a helicopter. It is critical though, that it is easy to return so that the user does not get lost during these excursions.
Obviously, there is a need for techniques which make the hypermedia structure comprehensible to the user. One approach is to visualize the contents and the structure of the system, and this is the technique we investigate in this thesis. However, there exist other techniques which can make the experience of the hypermedia space less abstract, such as input devices that provide tactile and force feedback (Brand 1988).
How disorientation problems can be reduced - visual communication
Visual communication is important for orientation because visuals give us so much spatial information. Consider a map, for instance. It pro-vides an excellent overall view (that is survey knowledge), which would be difficult to communicate in a non-visual way. It has been pointed out previously that spatial orientation devices could be particularly useful for helping a user build a cognitive map of a hypermedia document (Edward & Hardman 1989).
As an additional illustration, consider the following description of a house:
"Room A is located to the left of room B. Room B is above C. D is to the right of C."
It is not immediately clear in our mind how the rooms are located. By contrast look at this picture of the same house:
The locations of the different rooms become much clearer when visualized. However, it is likely that you first looked at the picture (since pictures strongly attract attention) and then read the text. You may want to try to visualize the house from the textual description while covering the picture.
Furthermore, visualizations can reduce the cognitive overhead of having to remem-ber the structure of the hypermedia network. It is a well known fact that humans are much better at recognizing than recalling information (Anderson 1985). Thus, visualizations can also be important for helping the user to recognize different locations and points of reference in a hypermedia structure, like landmarks or lighthouses. This has previously been pointed out by Nielsen (1990).
Communicating points of reference and giving an overview of a system using visualization techniques could be one way of addressing disorientation pro-blems. The burden of having to keep track of where we currently are and where we have been, could be eliminated by displaying that information right before our eyes. Devices like ever present maps (Apple 1989) could help in showing the context and maintaining a sense of position in the hypermedia space (for further discussion on maps, see chapter 5).
A visual presentation does not have to be in the form of a picture. As mentioned above, a text has visual properties which can function in a way similar to a pictorial presentations. Headings use a large type size, sub-headings one level smaller, and so on. This makes it easy to visually recognize key elements in the presentation. In addition, not all information lends itself to pictorial presentation. In some cases, when the subject can not be described pictorially, textual informa-tion can give a good overall view. For instance, consider a table of contents in a book. The layout makes it easy to find information. Moreover, we can use established search conventions, like the alphabetical search, when scanning an index or reference listing.
An important question is how to visualize orientational information. That is the topic of the next chapter.