Improving the Legibility of Virtual EnvironmentsRob Ingram and Steve BenfordDepartment of Computer ScienceThe University of NottinghamNottingham, NG7 2RD, UKTel.: +44 602 514203Email: s.benford, r.ingram @cs.nott.ac.ukFax: +44 602 514254http://www.crg.cs.nott.ac.uk/1. IntroductionYears of research into hyper-media systems have shown that finding one’s way throughlarge electronic information systems can be a difficult task. Our experiences with virtualreality suggest that users will also suffer from the commonly experienced “lost in hyper-space” problem when trying to navigate virtual environments.The goal of this paper is to propose and demonstrate a technique which is currentlyunder development with the aim of overcoming this problem. Our approach is based upon theconcept of legibility, adapted from the discipline of city planning. The legibility of an urbanenvironment refers to the ease with which its inhabitants can develop a cognitive map over aperiod of time and so orientate themselves within it and navigate through it [Lynch60].Research into this topic since the 1960s has argued that, by carefully designing key features ofurban environments planners can significantly influence their legibility.We propose that these legibility features might be adapted and applied to the design ofa wide variety of virtual environments and that, when combined with other navigational aidssuch as the trails, tours and signposts of the hyper-media world, might greatly enhancepeople’s ability to navigate them. In particular, the primary role of legibility would be to helpusers to navigate more easily as a result of experiencing a world for some time (hence the ideaof building a cognitive map). Thus, we would see our technique being of most benefit whenapplied to long term, persistent and slowly evolving virtual environments. Furthermore, weare particularly interested in the automatic application of legibility techniques to informationvisualisations as opposed to their relatively straight forward application to simulations of thereal-word. Thus, a typical future application of our work might be in enhancing visualisationsof large information systems such the World Wide Web.Section 2 of this paper summarises the concept of legibility as used in the domain ofcity planning and introduces some of the key features that have been adapted and applied inour work. Section 3 then describes in detail the set of algorithms and techniques which arebeing developed for the automatic creation or enhancement of these features within virtualdata spaces. Next, section 4 presents two example applications based on two different kinds ofvirtual data space. Finally, section 5 presents some initial reflections on this work anddiscusses the next steps in its evolution.2. What is legibility?Legibility, in the context of navigation and wayfinding, is a term which has been usedfor many years in the discipline of City Planning. Work on legibility in this area has beenconcerned with the way in which people are able to ‘read’ an environment and hence performwayfinding tasks. In his book “The Image of the City” [Lynch60] Kevin Lynch defines thelegibility of a city as: “…the ease with which its parts may be recognised and can be organisedinto a coherent pattern…” Here, Lynch is referring to the formation of a cognitive map withinthe persons mind [Passini92], a structure which is an internal representation of an environmentwhich its inhabitants use as a reference when navigating to a destination. The Image of theCity describes experiments carried out in a number of major US cities which show how thecognitive map is built up over time through experience of the city. The experiments involvedobtaining information from long term inhabitants of the cities in the form of, for example,interviews, written descriptions of journeys through the city and drawn maps. By examiningthis data Lynch identified five major elements of urban landscapes which are identified by theinhabitants and used as the building blocks of the cognitive maps. These features are:• Landmarks. Static and recognisable objects which can be usedto give a sense of location and bearing• Districts. Sections of the environment which have a distinctcharacter which provides coherence, allowing the whole to beviewed as a single entity• Paths. Major avenues of travel through the environment such asmajor roads or footpaths• Nodes. Important points of interest along paths, e.g. roadjunctions or town squares• Edges. Structures or features providing borders to districts orlinear obstacles3. Legibility techniques for virtual environmentsThe aim of ongoing research at Nottingham University is to apply the work describedabove not to real environments but to the artificial spaces of virtual reality systems. Morespecifically we are developing techniques to automatically construct the five legibility featuresin the abstract spaces produced by data visualisation systems such as database or documentstore visualisers. One of our main aims is to accomplish this with without requiring the usersof the system to perform the placement of the features manually. Essentially the systemshould, wherever possible, identify and place the features using information available from thedatabase and visualisation systems alone.To do this we have constructed a prototype system called LEADS (LEgibility forAbstract Data Spaces) which is designed to provide a layer on top of existing visualisationsystems and which performs the addition of legibility information to the space. LEADS actson the position of data items provided by this underlying system, as well as accessing the rawdata where necessary, to add and emphasise the objects which are used to improve thelegibility of the environment.LEADS is designed to be applied to spaces which satisfy three main criteria. Theyshould: be persistent over relatively long periods of time; be relatively stable so that theyevolve over their lifetime and are rarely disturbed by major upheavals in the database; beaccessed repeatedly by a number of independent users. An example of a large space to whichapplication of LEADS techniques might be appropriate is the WWW space, which isconstantly evolving but which rarely undergoes global scale restructuring.To place the legibility features LEADS uses districts as a starting point as this allowsfor a number of relatively simple techniques to be used to form the other features. Districts areareas of