Spatial and Textual Structures in Virtual Reality
David Modjeska
Department of Informatics
Umee Universiy
901-87 Umee
Sweden
+46 90 786 6825
dmjeska@informatik.umu.se
INTRODUCTION
Navigation in virtual worlds is a relatively new phenomenon. Although some behavioral studies have considered user navigation in this context, most existing literature discusses navigation in the physical world or hypermedia. Relevant physical research has been done in the fields of cognitive anthropology, cognitive psychology, and urban design. Given the abstract nature of information domains, a deep tension arises between semantic and physical structures. These structures have radically different properties. Accordingly, it is essential for the designers of virtual worlds to understand the tradeoffs between these types of structure, particularly for the important task of user navigation.
A recent CHI workshop suggested that navigation has two components: locomotion (an activity at the articulatory level) and wayfinding (an activity at the semantic level). With regard to wayfinding in the physical world, Passini proposed that this activity comprises three iterative stages: mental mapping, plan development, and plan execution. For mental mapping in the physical world, Lynch argued that key structural elements in urban design can enhance the legibility (or ease of mental mapping) of a city. Rennison and Strausfeld, finally, suggested that Lynch's structures can be mapped directly to computer information structures and information design attributes. Our current research seeks to examine the effects of variations of a fundamental design parameter of virtual worlds - the trade-off between textual and spatial structures - on a user's mental mapping process.
In assessing these design tradeoffs, the first important issue is how users and designers perceive trade-offs between textual and spatial structures. It is important for research to establish some reference points in the design space, where users and designers can agree on the relative importance of particular structures.
The second important issue is the user's experience of the virtual world - was it satisfying? - and his success in learning domain structure. A positive user experience will probably correlate with successful navigation. Subjective user feedback, moreover, provides the research with information about the qualitative properties of the experimental worlds. Furthermore, electronic-world navigation involves two basic tasks - searching and browsing. Search-engine technology and availability have provided effective solutions for the user's searching tasks. But techniques to support browsing lag behind. In browsing tasks, informally speaking, a user seeks to determine the nature and structure of available information. Accordingly, a user's success in learning domain structure during navigation is an important criterion for evaluating the success of a virtual world design.
The third important issue is the effectiveness of different virtual-world designs in representing and inculcating particular information structures. Urban-planner Lynch developed a concise and complete set of physical structures (path, edge, district, node, and landmark), which computer scientists Rennison and Strausfeld correlated with basic information structures (graph, linear, set, radial, and relational); information designer Wurman further correlates these structures with common information attributes (category, hierarchy, geography, continuum, and alphabetical). Since a virtual-world design is likely to support only some information structures effectively, it is important to examine this correlation empirically. In addition, the Lynchian design elements provide a useful analytical framework for assessing the structural domain learning discussed above.