[Contents]--[Abstract]--[1]--[2]--[3]--[4]--[5]--[6]--[7]--[8]--[References]
Metaphors allow users to reuse knowledge about objects in daily life in the virtual world. Metaphors do not represent the real world perfectly, instead they omit various features and provide others, new ones. The virtual world therefore can be an "improved" version of the real world -- at least in respect to the task at hand.
A reason for this trend towards spatialization is the amount of information people have to keep organized in their computer systems - a flat organizing scheme seems not to be sufficient and adequate for the great number of data objects users handle today. People are used to organize their lives spatially - be it at home or in the office place. The world we live in is a space. It therefore seems only logical that our computer systems are spaces the more they get realistic.
Such information spaces allows users to move things closer or further away without hiding those objects. They allow the user to move around in the virtual world and to shift her focus from one set of objects, or work-context, to another. This way to handle objects is closer to real office-work than the metaphor of the desktop, where all objects seemingly have the same distance to the user and the user never changes context.
As a spatial user interface is an environment this process of changing focus is called "navigation". We know how to navigate in environments we live in but we have little experience in navigating the virtual worlds of spatial user interfaces. As far as the spatial user interfaces use metaphors, that stay very close to the real world our navigational knowledge may be sufficient also to handle navigation in virtual worlds.
When enhancing virtual worlds with navigational features, that are not metaphors of the real world, our navigational knowledge is not sufficient any more and navigational problems occur. Designers of such enhanced virtual worlds therefore are facing many open questions like: How shall users move? How big should virtual worlds be? How shall objects be represented? Do we need or want all information to be visible? How can people interact in such spaces, how can they share information? How can these spaces be made really useful virtual worlds?
Hypertext
Navigation is also a process in environments, that are not based on real world metaphors. Every information seeking task in a computer system is a navigational process.
A general concept of an information space is the hypertextual view of information. Hypertext (or hypermedia) is an extension of the linear or strictly hierarchical structure of information we commonly use to form text documents, books and so forth. Texts like this thesis normally are read in a certain sequence - from front to back. If the reader skips over parts of the text she could miss important information in this intentionally laid-out structure.
Hypertexts present information differently. They have a web-like structure where various pieces of information, called "nodes", are connected by "links". Hypertext is often described as being similar to a dictionary where bits of information can be read in isolation but refer to each other. When reading a dictionary most readers soon start leafing forth and back, looking at cross-references. The basic concept is indeed the same, however hypertext runs on a computer and is interactive. Following a cross-reference in a dictionary needs time - in hypertext the corresponding information is displayed almost instantaneously. The fast linking process in hypertext allows users to read the hypertext "associatively" in ways impossible in a lexicon. Computer based hypertext systems further can make use of dynamic media like sound, animation, or video. This multimedia variant of hypertext is often called hypermedia. In the context of this thesis I will use hypertext and hypermedia synonymous.
As hypertext uses not a hierarchical or linear structure for the information space people sometimes have problems navigating the system. Many hypertexts suffer from the fact that it is hard to find useful information efficiently. Related problems are re-finding of information, and how to find out quickly that information looked for is not contained in the system at all.
Another, related problem is disorientation in the hypertext. This is sometimes called the "lost in hyperspace" problem. Recent research as discussed in a ECHT'92 panel indicates that the "lost in hyperspace problem" is a problem of communicating the structure of the information domain clearly to the user (see [BROW91]). Navigation in hypertexts therefore is not a problem inherent to hypertext but a problem of the interface between the user and the computer.
The task of navigating an information space occurs over and over again in the computer domain. Like in the case of hypertext navigation problems are inherently a problem of communicating the structure of the information space to the user and therefore a user interface problem. Navigation is easier when using a clear spatial user interface metaphor in the user interface, as this metaphor makes the spatial concept clear to the user and users are able to transfer more navigational knowledge from their daily experiences into the virtual world.
Virtual Reality
Virtual reality systems are a novel technology which could change our lives in the future enormously as it could provide the ideal technology for the realization of spatial user interfaces. In the virtual reality system the user is immersed in the virtual word and can interact with objects in a much more direct way than in current user interfaces.
Most virtual environments are relatively simple and small because of the computer resources needed to realize them. As computers get more powerful those environments will grow faster than our skills in designing navigational support for them. Although the spatial environment is meant to support navigation, these systems again could be hard to navigate because of the sizes they will acquire.
Learning from "world designers"
People spend their lives in the spatial environments they live in - be it cities, villages or buildings and cope well with the complex task of reaching their working places or homes every day. Cities and countries are environments, that allow us to navigate effectively in a large space. Architects and city planners realized a long time ago that way-finding and "navigation" in those environments can be made easier.
People like Kevin Lynch, Christopher Alexander and many others worked and still work to develop principles of how to make an environment pleasant and easy to navigate. Besides those ready available principles there is a lot of navigational infrastructure in real world spatial constructs like cities which can be adapted and reused in virtual environments. This infrastructure provides travelers with a large amount of navigational information and navigational support. Examples for such infrastructure are signs, traffic signs, information kiosks, transport systems and - last but not least - other travelers to communicate with.
Spatial user interfaces using interface metaphors based on these real life constructs can profit enormously from experiences in designing real-life spaces and the ways navigational infrastructure in the real world works. Indeed most virtual reality systems stay close to concepts of real life environments.
Textual Virtual Environments
While immersive virtual reality systems still tend to be small spaces there is a kind of virtual reality that commonly uses very large spaces already today: the text based virtual environment. Examples of such systems are networked adventure games like Multi User Dungeons (or MUDs). This systems realize today many of the features future virtual reality systems will show, but use descriptions of the environment instead of graphical representations.
MUD systems commonly use the spatial metaphor of a city or a landscape. They differ from most other current virtual environments in the fact that they are designed to be used by many participants at the same time and that they are much larger than current graphical virtual environments. They are large enough to observe the problems of navigating in large virtual environments. They also implement many navigational features that are not based on real-world metaphors.
Reviewing literature about spatial perception and memory, architecture and city-planning essential basic knowledge from these fields can be provided for spatial user interface design. Knowledge from the architectural domain is used to get an understanding of design issues for spatial environments. A novel spatial user interface metaphor, called the "Information City" is described, which is based on this knowledge.
To get a deeper understanding of navigation in virtual environments a study about navigation in text based virtual environments is described. This knowledge is set in relation to the principles of architecture and city-planning. This combined approach results in design knowledge and ideas, that can directly be used to implement a system based on the Information City metaphor in a textual virtual environment.
Expected results
The study described in this thesis allows interpretation of navigational strategies and experiences in context of general knowledge about environmental design gained in the literature review. Most virtual environments provide navigational features that assume a different concept of space than environments we are used to. When encountering such features in our daily lives, for instance teleporting, we would see them as "magic". In the virtual environments these "magic features" provide users with novel means of navigation and therefore are assumed to be useful. Parts of the study focus on magic features to find out if they really make navigation easier and how users use them.
The Information City metaphor relies on magic features to be a useful environment. The knowledge gained about how to design magic features, where to use them and where not to will make it possible to improve the concept of the Information City and to implement it in a textual virtual environment.
Hypertext
The hypertext field uses the concept of the "node" for a piece of information in the hypertextual information space. Nodes are interlinked using "links". Links lead from a "link anchor" in one node to a "link destination" which is either another node or a part of a node. Link anchors often are made visible by underlining a word or by using a symbol. This visible form sometimes gives information on the nature of the link destination. In this case the link is said to be a "typed" link.
A "path" is a sequence of nodes and links the user follows to reach a certain node. This concept of the path is similar to the concept of access-paths used in computer systems to access documents. The "history list" shows a list of recently visited nodes in the hypertext. It is a path-like concept since "looking back at the path" shows a list of nodes encountered in temporal ordering.
Hypertext nodes the user knows very well and which are recognized easily are called "landmarks". Examples are nodes containing initial screens of a system or tables of contents. When encountering a landmark the user knows where she is, because landmarks always help in orientation. Hypertext landmarks often contain helpful information or links to helpful information.
This summary shows that although hypertext normally uses no spatial concept it still makes use of environmental terms.
Real environments
In a real, spatial environment the concepts of node, path and landmark have a slightly different meaning. This ontology stays close to the terms defined by Kevin Lynch in his book "The image of the city" about the navigational task in a city [LYNC60]. He called the mental representation of an environment the "environmental image". It consists of five elements: node, path, edge, district and landmark.
A "node" is a distinct location in the environment. This may be a place, a piece of a forest path, a room or whatever - it is a strategic spot in an environment the user can enter. This description shows a first difference between the node concept in hypertexts and environments: as most hypertexts use no explicit spatial concept a hypertext node cannot be entered in the spatial sense.
Another element of the environmental image is the "path". It is a channel along which the observer moves. For many people paths are the dominant environmental elements as people often remember spatial concepts in terms of paths. Paths connect nodes or lead to nodes.
Another linear element of the environment is the "edge". Edges are borders perceived in the environment. The perception of a linear element as border or path depends mainly on the perspective of the observer - what is a path for the one person may well be seen as edge by another person. For instance what is seen as path by a car driver is seen as edge by a pedestrian trying to cross the street.
"Districts" are sections of cities perceived as one area because the objects (buildings) show common character. That common character can be functionality provided in the area, like in a harbor area, or the age of houses, like in the center of old cities. Districts often are surrounded by paths or edges.
The last of the five elements is the "landmark". Landmarks are strong points of reference often to be seen from far off.
Virtual environments
The differences between hypertexts and city environments are easy to grasp because hypertext does not show a clear spatial concept. Real environments and virtual environments are not that different. They both use a spatial concept and the main differences are in the existence of "magic features" in the virtual environment.
Textual virtual environments do not use the concept of the node for a location but the notion of the "room". The connections between rooms are called "exits" although they are conceptually very similar to the hypertext links. However, due to the overall spatial concept, exits show a "direction", like "north" or "west". Sometimes special directions are used to describe the destination of the exit. These exits often are exits using "magic features".
Like in real environments a "path" in a textual virtual environment is a linear element in the environmental image. People do not remember paths visually but they remember the exits taken to walk a path. They do not look at the rooms they move through - MUD paths seem to consist only of exits.
In the textual virtual environment the user reads textual descriptions of rooms. These descriptions can describe a room in a building or a forest path. There is no outside of a MUD room and therefore the concepts of the edge and the landmark are not existing in the same form in the textual virtual environment. A landmark in a textual virtual environment therefore is simply a room, that is easy to recognize. Similarly the edge can be perceived only by reading descriptions of it in several rooms. It is constructed in the head of the user and does not exit as element of the virtual environment.
This description of terms in the virtual environment applies mainly to the textual virtual environment. Indeed a graphical virtual environment - for instance a virtual reality system - lies in between the textual virtual environment and the real environment. Because of the graphical representation of objects and spaces containing them it is possible to create landmarks and edges that can be seen from far off.
Chapter 3 looks at environments people are used to navigate. The navigational task in cities is described and Lynch's work on navigation in city environments is reviewed. Starting from these foundations other aspects of navigation in real environments are reviewed like the use of space as an ordering principle, or communicating about space. The chapter concludes with a detailed ontology of the city environment.
Chapter 4 introduces user interfaces. First general user interface principles are reviewed with a special focus on the issue of user interface metaphors as a mapping from a source domain into a target domain. Spatial user interfaces are introduced and existing work in this very active field is reviewed.
Chapter 5 focuses at the navigation problem in user interfaces and virtual environments and describes the Information City metaphor.
Chapter 6 describes a study on navigation in textual virtual environments. First previous research in textual virtual environments and then the system used for the navigational study is outlined. The study consists of two parts. The methodology and the results are described in this chapter.
Chapter 7 contains the interpretation of the data gained in the study. The interpretation is based on examples taken from the virtual environment the study was performed in. The interpretation focuses on those aspects of the study that are directly applicable for the implementation of an Information City in a textual virtual environment.
Chapter 8 summarizes the main results of the thesis and gives a set of design guidelines for spatial user interfaces which could be generalized from the results of the study. Several areas of promising future research are pointed out.