Map-Based Navigation in a Graphical MOO - CiteSeerX

VIII. Appendix A – Paper submitted to Designing Interactive Systems, December 2001.

Map-Based Navigation in a Graphical MOO Wendy A. Schafer, Doug A. Bowman, John M. Carroll Virginia Polytechnic Institute and State University ABSTRACT Traditional MUDs and MOOs lack support for global awareness and simple navigation. These problems can be addressed by the introduction of a map-based navigation tool. In this paper we report on the design and evaluation of such a tool for MOOsburg, a graphical 2D MOO based on the town of Blacksburg, Virginia. The tool supports exploration and place-based tasks in the MOO. It also allows navigation of a large-scale map and encourages users to develop survey knowledge of the town. An evaluation revealed some initial usability problems with our prototype and suggested new design ideas that may better support users. Using these results, the lessons learned about map-based navigation are presented. Keywords map-based navigation, collaborative virtual environments, awareness, fisheye views, focus+context techniques INTRODUCTION Traditional MUDs and MOOs provide limited interaction and navigation. In a MUD (multi-user domain) or MOO (MUD object-oriented), users interact with one another and with the objects in the environment through simple textual commands. The interface is a scrolling sequence of commands and descriptions that pertain to the activities occurring in a particular room of the environment. Actions such as a user entering/leaving the room, speaking in the room, and picking up an object are displayed in the sequence. Movement through the world is accomplished with other textual commands such as 'go north'. Users navigate from room to adjacent room using their knowledge about the layout of the environment. This knowledge is acquired through repeated interaction with the environment, so that frequent users have a good mental model of the environment and new users have no knowledge of the spatial layout. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires specific permission and/or a fee. DIS2002, London © Copyright 2002 ACM 1-58113-2-9-0/00/0008 $5.00

It is particularly difficult for these interfaces to provide awareness information. The scrolling sequence of activity provides the user with awareness about a particular room, but does not provide any information about the other users or activities occurring in the environment. Users only learn about activities in a room if they enter that room by chance. This prevents users from finding interesting objects and activities in the environment and it limits the opportunities for collaboration between users. Traditional interfaces also limit a user's navigation of the environment. With most of the movement commands pertaining to an adjacent room, navigation through the environment occurs through multiple, repeated commands. This step-by-step process is not only tedious, but it also requires the user to recall the spatial layout in order to navigate between distant rooms. The layout is based on a directed graph such that going north and then east is not necessarily the same as going east and then north. Users must develop a mental model of this layout through repeated use. In gaming environments, this learning of the spatial layout may be part of the fun. But in MOOs primarily used for interacting with objects and other users, it may distract the user from other more relevant tasks. Map-based navigation provides one solution to the problems of traditional MUDs and MOOs. In map-based navigation, the text-based interface is combined with a graphical map of the environment. In this way, users do not have to develop a mental model of the spatial layout nor remember it. Also, the map can provide direct access to rooms through a clickable interface, eliminating the tedious movements. In addition, global awareness information can be displayed on the map. The map provides a graphical overview of the environment, which can be designed to provide awareness cues visually. For example, the map can indicate rooms with many objects, rooms with many users, or rooms that are visited most often. This should guide users to find interesting activities and encourage collaboration between users.

Map-based navigation also preserves the spatial aspect of traditional MOOs. The map interface represents the environment’s space and the rooms are placed relative to one another. Other interface widgets such as menus and search features would not support this spatial dimension. In particular, when the environment corresponds to a real area, the map establishes a connection between the real area and the virtual environment. The rooms become representations of physical places and their location within the space and with respect to another is important. Furthermore, when the target audience corresponds to local people in the real area, the map provides a familiar view of the location and can encourage a sense of community among the users. This paper addresses some of the design challenges introduced by map-based navigation. It describes the design of a map-based navigation tool for MOOsburg, a graphical MOO based on the town of Blacksburg, Virginia. This design supports the navigation of largescale maps with detailed information through fisheye views and a continuous zoom feature. The design also supports a user's mental representation of Blacksburg and it encourages corrections in this representation. Lastly, the design addresses issues related to designing maps for diverse user groups. RELATED WORK Thorndyke and Hayes-Roth describe three types of spatial knowledge (Thorndyke and Hayes-Roth 1982). Landmark knowledge is a familiarity with the visual attributes of an environment; Route knowledge is a familiarity with paths through an environment; and survey knowledge is a familiarity with an overview layout of an environment, similar to a map. People who develop a cognitive map of an environment through first-hand active navigation are likely to have an orientation-dependent map representation. Users of MOOsburg will be familiar with traveling through town as opposed to looking at a northup map of town. This implies that users will have greater landmark knowledge and route knowledge than survey knowledge. Lynch proposed that people use paths (e.g. walkways and passages), edges (e.g. walls and fences), landmarks (e.g. church spire), nodes (objects with similar characteristics), and districts (distinct sections) to build their cognitive maps (Lynch 1960). By placing these objects on a map, the map will better align with the users' conception of a space and better support the formation of spatial knowledge. For example, by including major landmarks in Blacksburg on our map, users will be able to make a connection between landmarks in their mental representation and the landmarks on the map interface.

Map-based navigation has been used in virtual environments (Bowman, Wineman, Hodges, and Allison 1999). Working with a north-up map, users can drag an icon of themselves from one location to another resulting in animated travel through the environment. Maps also support search tasks in virtual environments because they provide the user with survey knowledge that may not be learned through navigating the environment (Darken and Sibert 1996). Map-based navigation of a MOO differs in that the map is used to navigate directly and is not just an aid for navigation. Fisheye views use focus+context techniques to present the details needed for local interaction, but also include a compressed view of the overall structure (Furnas 1982; Lamping, Rao, and Pirolli 1995). Sarkar and Brown present a way to implement fisheye views for graphs using geometric transformations (Sarkar and Brown 1992). In their work, they suggest a specialized transformation for maps to make the fisheye view more natural. This transformation is different from others in that it not only makes items in the focus larger but it projects the map onto a hemisphere. We explore several projections in our map-based navigation tool. MAP-BASED NAVIGATION TOOL MOOsburg MOOsburg explores the use of a graphical MOO as a community application (Carroll, Rosson, Isenhour, Ganoe, Dunlap, Fogarty, Schafer, and Van Metre 2001). Most MUDs and MOOs are based on fictional places and are intended for almost any user group. MOOsburg, on the other hand, is based on a real place, the town of Blacksburg, Virginia. Its target user group is the town populace and the overall objective is to support community development within the town. These unique characteristics make map-based navigation a particularly good interface solution for MOOsburg. Because MOOsburg is based on an actual place, maps already exist for that place and we can reuse these maps in the interface. In addition, we can tailor the design to community members and make use of community knowledge such as major town landmarks. The structure of MOOsburg is also different than traditional MOOs. Instead of a network of rooms based on a directed graph, MOOsburg consists of spaces and landmarks, which correspond to points in a Cartesian space. Spaces are places in town that contain other places. Landmarks are visitable places in the MOO. A user either navigates to a landmark, or, if the landmark is itself a space with a substructure, the user can go into the landmark. Buildings are typical spaces and rooms are typical landmarks within these spaces. The map needs to support this hierarchy, as users will explore both spaces and landmarks. Figure 1 shows the MOOsburg interface including the map tool prototype.

about the landmarks. A red spot indicates the user’s current location on the map. This is similar to the “You Are Here” marker on public map displays. Spots colored light blue indicate the presence of other users, while dark blue spots correspond to places currently without visitors. We also represent the number of users at a location through the size of a spot, where larger circles correspond to more visitors. Lastly, spaces with substructure are drawn as unfilled circles and landmarks use filled circles. These visual cues provide awareness information about the various locations within the environment and guide users in the exploration scenario. The second primary task occurs when a user logs into MOOsburg with a particular objective. This objective may be related to a specific place, such as visiting the Virtual Science Fair at the Blacksburg Middle School, or it may be more general, such as learning the locations of all the coffee shops in town. In the Virtual Science Fair example, a user would work with the map to locate the school and then enter the school space to find the science fair. In the coffee shop example, the user would execute a search function that would cause the map to highlight all the places to get coffee. Both examples are goal-based scenarios. Figure 1. MOOsburg interface, including the map tool (lower right). User 'wschafer' is visiting the Drillfield landmark. MOOsburg supports community development by providing access to local information and to new collaboration activities. The project incorporates end-user programming to encourage relevant content and community involvement. In particular, users can add new landmarks and spaces to the MOO as well as develop new objects that provide collaboration opportunities. Usage scenarios Based on this description, the usage scenarios for MOOsburg are fairly different from those of traditional MUDs and MOOs. They focus on users finding information and collaborating with other users, whereas traditional MOO scenarios often focus on users navigating the environment and learning the spatial layout. This difference reflects the need for a different interface. In particular, the MOOsburg's interface should support efficient navigation and provide awareness information. Map-based navigation fulfills these requirements. We anticipate three primary types of tasks in MOOsburg exploration, goal-based tasks, and developer tasks. Many users will simply want to explore the MOO. This is the expected scenario for novice users and users just looking for interesting activities. For example, a new user will investigate the different spaces in MOOsburg and visit many landmarks. To encourage exploration, the map uses different visualization techniques to provide information

Lastly, the map interface needs to support end-user programming tasks. In particular, users need to be able to add spaces and landmarks to the map as they add content to the MOO. Our current prototype focuses on exploration tasks and place specific, goal-based tasks. A search feature will be investigated in the next version of the prototype to support other goal-based tasks. Also, providing support for end-user programming was not a priority in the prototype, as MOOsburg was still under development itself at the time. Map Prototype Our prototype tool provides map-based navigation in MOOsburg and allows us to investigate the usability of different map features. The prototype provides an accurate representation of the town through the use of digital data acquired from town authorities. The vector format of this data enables a Geographic Information System (GIS) style display and interaction techniques. Stored in the form of layers, the data allows either a single layer or multiple layers to be displayed. For example, we can display the road layer at certain zoom levels and both the building and road layers for more zoomed-in views. Each layer of data consists of a set of vector coordinates for the geometry of the objects in that layer. In this way, we can apply mathematical functions (Arc Tan, Hyperbolic, Parabolic x2, and Parabolic x1.5) to the coordinates to produce different projections. In the

map prototype, the different fisheye views correspond to different projections (see figure 3). The format of the data also allows a continuous zoom feature because it is vector-based. The map supports three main types of interaction clicking, dragging, and zooming. When users click on the interface, the map is redrawn with the location of the click placed in the centre of the map. Dragging the map produces a panning motion, revealing parts of the map not previously visible. Lastly, users can change the zoom factor of the map continuously through a slider widget. Figures 2 and 3 illustrate these map interactions.

Figure 2. The effect of zooming with the map prototype in the flat view.

Figure 3. The effect of clicking with the map prototype in the Parabolic x2 view. The map in MOOsburg needs to address the transformation from a mental representation of Blacksburg to an actual map. People need to recognize the map as the town of Blacksburg and be able to relate to it. We hypothesized that including major landmarks in town on the map would allow users to make a connection between places in their mental representation and major town landmarks we provide on the map. We can also support route knowledge, by including the roads and pathways of Blacksburg. However, a map that can be rotated might better support the users' orientationdependent spatial knowledge (Darken and Cevik 1999). Lastly, as the map displays survey knowledge of the town this may cause users to correct any inconsistencies with their mental representation. In working with the map, users will gain a better understanding of the layout of the town. This knowledge will be helpful as users continue to

interact with the map and it may cause users to develop a more accurate cognitive map of Blacksburg. FORMATIVE EVALUATION Following the implementation of the map prototype, we conducted a formative evaluation. This evaluation revealed some initial usability problems with our prototype and introduced some new design ideas that may better support users. Subjects The evaluation of the map prototype involved subjects from three different user groups - middle school students, college students, and seniors. MOOsburg is a community application and so it is important to have a variety of community members involved in its evaluation. We wanted to avoid an interface design that is useful for college students and yet causes problems for middle school students and seniors. In particular, we were uncertain about how well the middle school students would be able to work with a map of the town, since at this age map skills and spatial knowledge may not be well developed. Also, we expected the evaluation to reveal differences in the user groups and indicate if different user groups would benefit from different interface designs. We met with four users from each user group for a total of twelve subjects. Within each group, two of the users were female and the other two were male. The average ages of the middle school students, college students, and seniors were fourteen, twenty-three, and seventy-seven, respectively. All the users indicated that they use computers frequently to do a variety of tasks. Most of the college students indicated that they use map programs over the Internet once a month and that they consult atlases and other paper maps less frequently. None of the seniors or eighth graders had used Internet map programs previously, but both groups indicated monthly use of atlases and paper maps. Method Each user followed a similar four-step procedure that lasted about 20 minutes. The evaluation was intended to be think-aloud and users were asked to comment on their thoughts while performing the tasks. After filling out a background questionnaire, the users answered a series of questions related to flat map paper mockups. The first mockup was an unlabeled map of major roads in Blacksburg and we asked the user if he/she recognized the place depicted in the map. This was to investigate how familiar users were with the layout of Blacksburg and whether they could recognize it from just a sketch of the major roads. We also looked at the users' ability to work with less detailed, overview maps of the town and whether they

could understand one abstraction better than another. Ideally, users will be able to navigate to any place in MOOsburg if given an overview map of the town. We investigated what details need to be provided on this overview map to support navigation. To do this, we examined how successfully users pointed out places in town with three different maps - a map of major roads without labels, a map of major roads with labels, and a map with a set of possible major town landmarks labelled, in addition to major roads without labels. During this line of questioning, we noted road names and town landmarks the users mentioned. This allowed us to learn about the users' perceptions of the major roads and landmarks in town, where repeated use of certain roads and landmarks indicates their significance. We also observed how the users worked with maps. In particular, actions such as drawing on the map and rotating the map were noted. These actions would indicate that features such as editing or rotating capabilities might improve the usability of the map interface. In the second step, users worked with the map prototype to complete two tasks. The first task asked users to find and centre their home with no further instructions on how to work with the map. This allowed us to observe how easy it is to learn how to interact with the map and how intuitive the map interactions are. In the second task, we described for the users the three interaction techniques the prototype supports and asked the user to centre the downtown area. This gave the users an opportunity to experiment with the map further and try out all the features, while we again observed the intuitiveness of the interactions and the map's ease of use. In the third section of the evaluation, users were asked to point out four different places using four different fisheye view paper mockups. Each view showed the roads and buildings in one section of town. Observations were made about the users' initial reactions to the views in addition to how the users worked with the maps. Lastly, the users worked with the three fisheye views of the prototype. The users were asked to navigate to the places they pointed out in the previous step (this tested the goal-based navigation usage scenario). Two trials were run for each place where each involved a different view and different starting position. The first task compared two different fisheye views. In the second and third tasks the other two fisheye views were compared with the flat map view. Users were allowed to use the zoom feature for the second task but not the third. We collected the users' comments while working with the different views, their view preference after completing the tasks, and the time taken to complete the tasks. This allowed us to examine the usability advantages fisheye views may have over the flat map view.

RESULTS Section 1: Paper prototypes of a flat map We first observed the users working with different paper mockups of the town. When they were asked to identify Blacksburg from an overview map of major roads, only two of the eighth graders recognized the town, while three college students and all of the seniors answered correctly. After we informed the users that it was Blacksburg, some of the seniors and college students could point out their current location. Yet, none of the eighth graders could point out an approximate location for their current location, the Blacksburg Middle School. The school lies on the main road running through town, yet all of the eighth graders struggled to work with the map. When major road names were added, it was apparent that almost all of the users were familiar with the location of the grocery stores. Each senior pointed out two grocery stores when asked to point out where they shop. All of the college students were familiar with each grocery store's location. Three out of the four eighth graders knew what road their grocery store was located on, but only two of them could point it out. Next, the users were presented with a similar map with landmarks. Some of the college students were not familiar with all of the landmarks displayed, while all the seniors and the eighth graders indicated that they recognized the landmarks. For example, many of the college students could not identify with the different public schools on the map. All of the college students indicated that the labelled roads map was much more helpful than the landmark map (one user even requested to use the map with road labels to answer the question!). Similarly, many of the seniors indicated that the roads names were more helpful. During this section of the test, five of the twelve users were observed drawing on the paper mockups and at least one user from each group rotated a paper mockup. Section 2: Interacting with a flat view of the prototype Observing users experimenting with the map prototype revealed some key usability problems with the prototype. Most importantly, our observations showed that the clicking interaction technique is not intuitive. During first use, all of the seniors were confused by the tool's response to a click or they commented that clicking was confusing. Similarly, most of the college students used clicking just to move the map and did not realize that the map was recentreing where they had clicked. Two of the eighth graders did not seem to understand how clicking worked, while another indicated that it was confusing. With the senior users, it was apparent that some also struggled to operate the mouse. They often double-clicked in a location, causing them to move twice as far away as

While trying to modify the zoom level, many users were observed clicking on the labels 'Zoom In' and 'Zoom Out', clicking next to the slider, or clicking on the slider track. Also, at least one user attempted to use the zoom control to change the view as you would with a scrollbar. This is an understandable mistake, because the slider is conveniently located where a scrollbar would be. All of the eighth graders struggled to find their homes. They knew the general area where they lived, but it took some time for them to find it, inquiring about different buildings and roads on the map during the process. This suggests that younger, less-experienced users need additional cues in the form of labels and landmarks. Interestingly, two of the eighth graders did not recognize the Drillfield (a prominent grassy lawn in the shape of an oval on the Virginia Tech campus) and asked if it was a golf course. All of the seniors and college students easily pointed out this landmark and used it to orient themselves when working with the paper mockups and the map prototype. Section 3: Paper prototypes of fisheye-view maps As the users pointed out places with the different fisheye view paper mockups, we noticed that all of the users preferred the views that contained recognizable landmarks. We expected the users to comment about the advantages and disadvantages of the different projections, but instead they mentioned how important finding a landmark was to complete their task. We also noticed that if the users recognized a landmark on the map and they pointed out the wrong place, then they were usually close to the target. For example, some of the users pointed out buildings adjacent to the requested post office. The post office is in the heart of the downtown area with many buildings around it and it can be difficult to pick out. This indicates that more visual cues need to be provided to guide the user in the final stages of navigation. The seniors had many initial comments about fisheye views. These included: 'I'm buffaloed', 'This is really distorted', and 'Main Street is too elongated'. None of the users gave up in looking for the requested places, but these comments expressed their reservations about the fisheye views. Also, three of the four eighth graders could not point out a major intersection, not because of problems with the tool, but because they were not familiar with the name of the intersection.

Section 4: Navigation tasks involving fisheye views During section three, most of the seniors seemed to forget how to use the map tool. When given a navigation task in section four, they were unsure of how to move the map and had to relearn the interaction techniques. Also, two of the seniors also had to be prompted to zoom out in order to determine where they were located. Problems with clicking were apparent in this section of the evaluation as well. At least one senior continued struggled with the clicking interaction until the last task. Also, eighth graders were observed accidentally double clicking. This indicates that problems with this interaction technique are not limited to first-time use. When comparing a fisheye view to the flat view with zoom capabilities, most of the college students found the requested location faster with the flat view, where the largest time difference was 11 seconds. All of the seniors and all of the eighth graders completed the task faster with the flat view. Each of the eighth graders took about 25 seconds longer with the fisheye view, while two of the seniors really struggled with the distorted views. They took about twice as long to complete the fisheye task. As a result, figure 4 shows a larger time difference in the average results for the seniors.

300

Time (sec)

they wanted. Also, one senior struggled to drag the zoom slider and the map. Slow map response exacerbated these usability problems.

250

Flat View

200

Fisheye View

150 100 50 0

Eighth Graders

College Students

Seniors

Figure 4. Average navigation times using zooming, by user group and map type. When comparing a fisheye view to the flat view without zoom capabilities, most of the college students found the requested location faster with the fisheye view (avg. 16.75 sec) than with the flat view (avg. 54 sec). Likewise, most of the eighth graders completed the task faster with a fisheye view (avg. 32.25 sec), than the flat view (avg. 71.25 sec). Lastly, most of the seniors found the requested location faster with the fisheye view. The one exception involved an extreme difference where the user took four times as long, causing the group's time difference to reflect faster times with the flat view. On average, the

other seniors were actually 47 seconds faster with the fisheye view.

Time (sec)

These preliminary results suggest that when zooming is used, flat maps are better than fisheye views, and when zooming is not used, fisheye views are superior for most users. The seniors took more time to complete the tasks with zooming than the other users, suggesting that they struggled with multiple interactions. Also, the average times for the tasks with zooming are less than the average times in the other tasks for two user groups. This suggests that zoomable maps are more efficient for the eighth graders and college students in many tasks.

90 80 70 60 50 40 30 20 10 0

Flat View Fisheye View

Eighth Graders

College Students

Seniors

Figure 5. Average navigation times without zooming, by user group and map type. Strategies All of the users became more familiar with the map of Blacksburg as the session progressed. In the last couple of tasks, users could clearly identify the major landmarks they were using and in most cases the subjects were using them to navigate the prototype very efficiently. For example, at least three of the eighth graders were observed using the same navigation approach: identify Main Street, follow it to the Middle School, and then use the Middle School as a reference point. Using this approach during the navigation sections, the users repeatedly tried to identify the larger buildings in a view and use these buildings to orient themselves. Without building labels this was a challenge at times. Also, users occasionally had trouble because they identified a large building incorrectly. For example, one of the eighth graders became confused when he believed that a large building was the Middle School when it was not. Throughout the sessions, users also were observed adding to their set of well-known landmarks. For example, after pointing out the library in one task, users would recognize and use the library as a reference point in later tasks.

Lastly, many of the college students actually became disoriented when trying to orient themselves with the familiar Drillfield. The symmetric shape of the Drillfield causes this confusion, because it is hard to distinguish between two similar sides. DISCUSSION Map modifications These results directly indicate the modifications that we have made to increase the usability of the map tool. One simple change was to make the labels, slider track, and area next to the zoom control clickable, and move the zoom control to the left of the map. The map prototype also lacked a labelling scheme. In the sessions, we observed users trying to identify roads and buildings. This activity is now supported through both tool tips or fixed labels on the map. For the evaluation, the map only had two levels of detail all roads and, for more zoomed-in views, buildings and roads. The results indicate that when users navigate they use both roads and landmarks. The map was modified to display these guides at all zoom levels. At the most zoomed-out view, an overview map of major roads and major landmarks along with a labelling scheme will better support users (see Figure 6). Zooming in, the map displays increasing levels of detail. All roads are added to the overview map followed by all buildings at the most zoomed-in view. In this way, the map supports navigation tasks at all zoom levels. Lastly, changes were made to better support map interactions within MOOsburg. In particular, a click solution was implemented that makes mouse clicks intuitive and catches erroneous double clicks. Right clicking brings up a popup menu with an option to recentre the map. Choosing this option, the map’s movements are animated to provide better feedback to the user. Single clicks are reserved for selecting actual spaces and landmarks in the MOO. In this way, clicking on a marked space or landmark will cause the map to update, and the user's task focus will change from navigation to visiting the place selected. Lessons learned These results can be generalized into a set of suggestions for using map-based navigation in a multi-user environment. Some of these are especially applicable when the environment represents a real place familiar to users. First, individual users have different perceptions of the same place. In this study, we observed users with different mental representations of Blacksburg. For example, most of the eighth graders did not recognize a major intersection in town when we gave them the road names. They may have been familiar with this intersection, but known it by some other way, such as the intersection with the video store. In this way, the eighth

graders had a different perception of the intersection and, in turn, Blacksburg.

Figure 6. Modified map prototype based on evaluation results. The logos of local grocery stores provide orientation information to diverse user groups. Similarly, individual users are familiar with different areas of an environment. In the study, some of the users lived in the southeast section of town and were familiar with various places and roads in that area, while other users had little knowledge of that section of town. Also, when the users were asked to point out the post office and the library, they invariably indicated the post office or library in the part of town with which they were most familiar. In designing a map-based navigation tool, it is important to keep these individual differences, both different perceptions and different familiarities, in mind and support navigation of the entire map space. The evaluation also revealed that different user groups have knowledge of different landmarks in town. In the first section of the evaluation, most of the college students did not recognize the places thought to be major landmarks in town, while none of the eighth graders or the seniors had trouble with these landmarks. Also, the college students were constantly observed using the Drillfield to complete the tasks and yet two eighth graders inquired about this large oval shape during their sessions. This presents an interesting problem of what landmarks to use with a map-based navigation tool. If we include landmarks more familiar to one user group, they may cause the map to be less usable by the others. If we include landmarks familiar to all groups, the map may

become cluttered and less usable for everyone. Ideally, the map should include landmarks common to all groups. This evaluation revealed that grocery stores are a good choice. All ages visit grocery stores and they tend to be large buildings often located within noticeable shopping plazas. Designers of map-based navigation need to recognize these differences in landmark knowledge and support all user groups, possibly through a set of common landmarks, or by allowing users to select a customized set of landmarks. In addition to the differences in landmark knowledge, user groups differ in spatial knowledge and this difference is reflected in how they navigate. In our study, the eighth graders were observed to be more landmark-oriented than the other groups. For example, the eighth graders struggled with road-based tasks, such as pointing out a major intersection in town. Eighth graders do not drive and so they are less familiar with the roads in town than other users. We observed that the landmarks were as useful to 'passengers' as the road names were to the drivers. All of the drivers referred to road names throughout the tasks, while all of the eighth graders relied on the landmark knowledge. For example, during the evaluation one eighth grader mentioned that it was easier to navigate with a zoomed-in, roads and buildings view than a roads only view, even though this limited the area shown on the map. The buildings provided this user a primitive form of landmarks to use while navigating. This difference in spatial knowledge also applies to other grade school children who do not drive or have just recently started driving. A map-based navigation tool must take differences in spatial knowledge into account and support both the use of landmark knowledge and the use of route knowledge. A more general set of suggestions for supporting navigation can be also identified from the observations. First, recognizable landmarks are key for navigation tasks. When the users were asked to point out places on the paper mockups of the fisheye views, all of the users indicated a preference for the views in which they recognized a landmark. This further emphasizes the need for major landmarks to be included on the map. Similarly, while users navigated with the map prototype, they were observed looking for landmarks they recognized in order to orient themselves with the map. One user even commented: 'If you don't have a Drillfield, then you can't work'. This user was looking for the easily recognizable Drillfield and desired to use it to complete most of the navigation tasks. Not all of the users were successful in orienting with the Drillfield though. Some of the college students struggled with the symmetry of this landmark, thinking the north side was actually the south side and the east side was the west. These users needed more than the landmark, they needed clarification of its orientation. A

map-based navigation tool must include landmarks and display them so that they are recognizable and usable. Users are also drawn to large buildings. These buildings are prominent on maps and users will try to use them during navigation. During the evaluation, users would often point out a large building on the screen and ask about it. Again, users were trying to recognize a landmark and orient themselves with it. Designers of a map-based navigation tool need to be aware of the larger buildings in the environment and provide users with ways to identify these buildings. As each of the evaluation sessions progressed, it became clear that the users were learning. A uniquely shaped building, used as the target location in one task, was referenced by users in later tasks. Similarly, major roads pointed out earlier in the evaluation were referenced in later tasks. The users learned about other landmarks while working with the map and they reused this knowledge. This notion of learning should be kept in mind while designing a map-based navigation tool. One possibility for providing better support is to allow map annotations for future reference.

We are continuing this work by investigating other ways interactive map software can support collaboration. The map tool in MOOsburg provides a way to navigate a collaborative environment, yet map-based interfaces can also directly support spatial, collaboration tasks. For example, a group of people might be involved in planning a new park or learning about a local watershed together. In both examples, the focus of collaboration is a spatial object - an area of land and a network of waterways. Twodimensional and three-dimensional maps are a common way to represent spatial objects. We are exploring the use of both mapping techniques with respect to distributed collaboration. AUTHORS Wendy A. Schafer, Doug A. Bowman, and John M. Carroll Center for Human-Computer Interaction Department of Computer Science Virginia Polytechnic Institute and State University Blacksburg, VA 24061 USA Phone +1 540 231 7524 Fax +1 540 231 6075 {wschafer, bowman, carroll}@vt.edu

In addition to landmarks, users may also need visual reminders of how to navigate with the map. During the evaluation, none of the users used all three interaction techniques while completing the navigation tasks. In fact, most of the seniors had to be prompted to interact with the map when they were first asked to use it. They also had to relearn how to interact with the map after looking at paper mockups. In MOOsburg, visitable landmarks and spaces are marked on the map and they encourage users to click on them, but navigation controls also need to be apparent. The zoom control provides one type of affordance, but others are needed. A map-based navigation tool must provide visibility and affordances to support navigation interactions, possibly through the use of visible control widgets such as scrollbars (Norman 1990).

ACKNOWLEDGMENTS The authors wish to thank their subjects for their participation in the study. Other members of the MOOsburg group that were also helpful include Cara Struble, Craig Ganoe, Dan Dunlap, Philip Isenhour, and Dennis Neale. This project was possible through support provided by NSF grant DGE-9553458, the Office of Naval Research, and the Hitachi Foundation.

CONCLUSIONS AND FUTURE WORK Maps provide a good interface for place-based navigation. In MOOs, a map allows a user to browse and visit places while maintaining awareness. Our design of a map-based navigation tool for MOOsburg aims to support users navigating a familiar place.

2. Bowman, D. Wineman, J., Hodges, L., and Allison, D. The Educational Value of an Information-Rich Virtual Environment. Presence: Teleoperators and Virtual Environments, vol. 8, no. 3, June 1999, pp. 317-331.

Map-based navigation could also be used with web pages or in a team-based gaming environment. For example, if a collection of web pages dealt with specific places, a map would provide a good navigation tool. In a game, the map would support simple navigation between places and it would provide awareness information about team members and opponents.

REFERENCES 1. Carroll, J.M., Rosson, M.B., Isenhour, P.L., Ganoe, C.H., Dunlap, D.R., Fogarty, J., Schafer, W.A. and Van Metre, C.A. (2001). Designing Our Town: MOOsburg. Accepted for publication in International Journal of Human-Computer Studies.

3. Darken, R., and Cevik, H. Map Usage in Virtual Environments: Orientation Issues. Proceedings of Virtual Reality '99. 1999. IEEE. pp. 133-140. 4. Darken, R.P., and Sibert, J.L. Wayfinding Strategies and Behaviors in Large Virtual Worlds. Proceedings of Human Factors in Computing Systems. 1996. ACM. pp. 142-149. 5. Furnas, G. The fisheye view: A new look at structured files. Technical Memorandum 82-11221-22, Bell Laboratories, 1982.

6. Lamping, J., Rao, R., and Pirolli, P. A focus+context technique based on hyperbolic geometry for visualizing large hierarchies. Proceedings of Human factors in Computing Systems. 1995. ACM. pp. 401408.

8. Norman, D. The Design of Everyday Things. 1990, New York: Doubleday.

7. Lynch, K. The Image of the City. 1960, Cambridge: MIT Press.

Thorndyke, P., and Hayes-Roth, B. Differences in spatial knowledge obtained from maps and navigation. Cognitive Psychology, 1982. 14: pp. 560-589.

9. Sarkar, M., and Brown, M.H. Graphical Fisheye Views of Graphs. Proceedings of Human Factors in Computing Systems. 1992. ACM. pp. 83-91.

IX. Appendix B – Boaster presented at the Human-Computer Interaction Consortium in February 2001.

Using Interactive Maps in Community Applications Wendy A. Schafer Center for Human-Computer Interaction Department of Computer Science Virginia Tech Blacksburg, VA 24061 USA +1 540 231 7524 [email protected] ABSTRACT

homes, about a map of a nearby empty building lot.

Interactive maps provide unique ways to support community applications. In particular, they enable new collaborative activities. Map-based navigation supports a community environment as well as virtual tours. Interactive maps can also function as a tool in collecting historical information and discussing new spatial layouts. These examples indicate the numerous opportunities for interactive maps to support collaboration.

Maps are an ideal way to represent community space. People are familiar with looking at maps. Atlases are commonly found in people's cars, many people have looked at a globe, and even a trip to the mall has people glancing at a map. Also, maps already exist for many spaces. For example, both town maps and building floor plans are common occurrences. Most importantly, maps can exploit the common, community knowledge of the space they represent. By including well-known landmarks and highly visited places on a map, users will more easily identify with the representation and, in turn, find the map easier to work with.

Keywords Collaborative activities INTRODUCTION Communities are often associated with physical space. For example, a community of local people is usually associated with a town. Likewise, an office space has a certain community of workers. This space is the commonality between the community members and causes many community activities to be spatial. As a result, software for communities often requires a representation of this space. For example, a community web site usually includes a map. The map might be an imagemap with links to various areas of the space or the map might provide community information, such as interesting places. There are numerous uses for a community map on the web. Web browsers and most web pages are designed only as single-user applications, but communities can also benefit from spatial representations in collaborative applications. Such applications encourage community development as they can bring together people independent of distance and time. For example, a collection of neighbors might carry on an asynchronous discussion, from their own

Adding maps to an application's interface is not limited to just a static picture. Interactive maps provide new features over traditional maps and support novel interactions. For example, an interactive map could support zooming, a task that would usually require additional flat maps. In particular, vector-based map data provides many possibilities for interactivity. With this format, software programs can easily manipulate the data and provide different displays in real-time. For example, a user could interact with a map by adding and removing different layers, changing the display scale, and applying different projections. The use of interactive maps in community applications opens up a range of new activities. Particularly interesting are the new opportunities for collaboration. The rest of this paper highlights some of the ways interactive maps can be used to support collaboration in a community application. First, it discusses the use of map-based navigation in a collaborative environment. A map-based navigation prototype and its evaluation are described, along with some lessons learned and recent extensions. Other community activities that could be enhanced with interactive maps and collaboration include virtual tours, historical collection, and spatial planning. Ideas about each are discussed. MAP-BASED NAVIGATION

MOOsburg is a collaborative environment based on the town of Blacksburg, Virginia (Figure 1). The goal of this environment is to support community development in Blacksburg by providing community members with a new way to communicate. Ideally, users will build distinct locations within the environment, similar to rooms, and establish a community network [4]. Typically, the room metaphor implies that all the interactions that occur within a room are only viewable by the local users present. This allows for multiple, independent conversations to be going on at the same time, but it limits users’ awareness of one another. For example, a user will only learn about the interesting activities in a room if he/she enters that room by chance, limiting the opportunities for collaboration. Yet, distinct locations have many positive attributes. They can encourage the development of social networks and support individual and group activities, as well as synchronous and asynchronous work [2,3]. There is not a definitive design for how users navigate and explore this type of environment, but ideally we would want to limit isolation and support awareness. Traditional text-based MUDs and MOOs offer one way to implement distinct locations. Yet, this approach is not well suited for collaborative environments. Most of the movement commands pertain to an adjacent room and navigation through the environment occurs through multiple, repeated commands. This step-by-step process is not only tedious, but it also requires the user to recall the spatial layout. An alternative implementation involves map-based navigation. An interactive map provides a visible structure for the environment and enables direct access to locations. In this way, users can focus on the collaborative activities occurring within the rooms, rather than navigating between rooms.

For example, someone may log on to visit the Virtual Science Fair happening at the local middle school. Users navigate by interacting with the map through zooming, dragging, and clicking. The zoom level is controlled continuously by a slider widget. Users can also click and drag the map, causing a panning effect to occur, or click on the map to have the clicked location move to the center. The use of vector data also enables the prototype to support layers and projections. Major roads and major landmarks are displayed at zoomed out views. By zooming in, all roads are displayed followed by roads and buildings at the most zoomed in views. We also have implemented four different projections that provide a collection of fisheye views (Figure 2). These projections give the user extra contextual information on the periphery of the map when working with zoomed in views. The map-based navigation prototype, including these features, is available as a demo on the web at: http://java.cs.vt.edu/~wschafer/Mapview.html

Figure 2. Map-based navigation prototype displaying a fisheye view based on a parabolic function. Formative Evaluation Twelve users from three different user groups in the community helped guide our prototype design. Middle school students, college students, and senior citizens completed navigation tasks using both paper mockups and the map prototype. The results suggested some guidelines for map-based navigation.

Figure 1. MOOsburg interface, including the map-based navigation prototype (lower right). User "wschafer" is visiting the Drillfield location. Prototype Design Our map-based navigation prototype supports two typical scenarios of MOOsburg. It allows users to explore the environment and virtually visit various places in Blacksburg, and it supports specific place-based tasks.

First, we observed that individual users have different perceptions of the same location. For example, some of the users did not recognize a major intersection in town when we gave them the road names. Yet, they seemed to know this crossroads in a different way, such as the intersection with the video store on the corner. Users also differed in their familiarity with the environment. For example, those who lived in a particular section of town knew more about the various places and roads in that section than the other users. Differences also exist between user groups, particularly in respects to landmark knowledge and navigation strategies. For example, places believed to be major landmarks in

town were recognized by the middle school students and senior citizens, but not by the college students. In respects to navigation strategies, the college students and senior citizens would utilize road names throughout the session, while the middle school students relied solely on the buildings to navigate. This reveals a need for landmarks in all views, not just in zoomed in displays.

and encourage collaboration between users. Our prototype portrays the current user activity by using variable spot sizes, based on the number of users at a location (Figure 3).

One part of the evaluation asked the users to point out places on paper mockups of the fisheye views. During this task, all of the users indicated a preference for the views with recognizable landmarks, independent of the projections. Along the same lines, we also observed users trying to recognize large, prominent buildings. This emphasizes how important landmarks are to navigation. The evaluation also demonstrated that users easily learn the shapes of buildings and roads. As the sessions progressed, buildings and roads discovered in previous tasks were referenced in later tasks by all of the users. Lastly, we observed that users need visual reminders of how to interact with the map. Map-based navigation in a collaborative environment is a two-step process navigation to a place and then participating in an activity at that place. Many of our users forgot how to work with the map when we varied paper mockup tasks and prototype tasks. This indicates that users will need visual reminders of map interactions after visiting a place in a collaborative environment as well. Extensions Following the evaluation, the prototype was additionally enhanced to support MOOsburg's hierarchy of spaces (containers) and places (landmarks). A user either navigates to a landmark, or, if the landmark is itself a space with a substructure, the user can go into the landmark. Buildings are typical spaces and rooms are typical landmarks within these spaces. This allows users to model Blacksburg more accurately and it provides structure to the community network. Each space corresponds to a different map and as a user chooses to enter a space, the new map is displayed with the previous map shrunk to a small icon (Figure 3). The small maps remind users of their location within MOOsburg as they explore subspaces. They also provide an easy way to exit subspaces, where clicking on a small map returns you to that space. For example, clicking on the small map in Figure 3 would cause the full Blacksburg map to appear in the window and the user would return to the Blacksburg space. The prototype also has been extended to function as an awareness tool in MOOsburg. The map provides a graphical overview of the environment, which can be designed to provide awareness cues visually. For example, the map can indicate rooms with many objects, rooms with many users, or rooms that are visited most often. This should guide users to find interesting activities

Figure 3. Extensions to the prototype - support for a hierarchy of spaces and places and awareness information about the number of users at a location. OTHER APPLICATIONS Interactive maps can be used to support a range of collaborative, community activities. For example, in real space communities often offer tours to introduce nonlocals to their community. This is mimicked on the web as a virtual tour, where individual users often follow a step-by-step directed path. Yet, this experience could be greatly enhanced as a collaborative activity using a map. A tour would be more engaging if a user followed a dynamic route guided by a community member also using the system. Map-based navigation for a community environment could support this activity by allowing users to link their navigation and/or viewpoint of the map. Linking navigation would provide a way to synchronize travel through the environment, eliminating much discussion about where the tour was headed next and how to get there. Linking viewpoints would allow users to discuss structural layouts of the environment along the tour, as everyone would have an identical view of the map. Map-based navigation of a community environment could also enhance a single-user tour application. For example, the map could recognize common paths through the environment and suggest these to newcomers. This might prevent a user from getting an exhaustive tour, but it would indicate specific places where other community members visit and suggest the activities they participate in. This use of an interactive map is not necessarily collaborative, but it will encourage collaboration as the user taking the tour may find community members or activities that interest him/her. Another collaborative, community application enabled by interactive maps involves the collection of historical information. In real space, communities often have

records of their history, such as member lists, records of community events, and records of major community changes. This information is not always complete or accurate as often only a few people are in charge of advancing and maintaining this history. An interesting collaborative activity is to encourage community members to remember the past and document it in a community application. This could move the role of historian from a few people to a community-wide effort. If publicly available, it could also give outsiders a feel for how the town once was and how it has progressed. Interactive maps enable this type of activity, as spatial changes are an integral part of a community's history. In fact, older maps of the community could set the stage for such an activity and prompt users to remember the past. People enjoy reminiscing about "the way things used to be". For example, it is fun to remember previous roadways, buildings, and businesses in a town. An interactive map could be used as the tool to recreate this history. It could allow users to page through history, look at community member’s contributions, and add their own recollections. Another opportunity provided by interactive maps is a collaborative planning application. Communities are typically interested in new spatial plans. For example, townspeople like to evaluate how new roads and new buildings will affect their community and voice their concerns. Likewise, office workers want to be involved in deciding how a new open space will be laid out and partitioned. An interactive map directly supports these discussions. It can display different spatial layouts such as the current arrangement, different proposals, or different combinations superimposed. The map could also allow collaborative annotation so that the communication channel is extended beyond talking. For example, two small groups could have independent discussions involving many annotations before they gather together to compare their thoughts and prepare a single list of concerns.

User interactive maps within community applications, like historical collection and planning, is different than multiple users looking at a geographic information system (GIS). GIS systems are typically used to analyze spatial data sets. These systems focus on details, precision, and static, presentation views. For example, a GIS map could indicate an ideal park area by simultaneously displaying soil types and wildlife patterns for a region. Interactive maps, on the other hand, are designed so that casual users can work independently or collaboratively. Their use occurs at a high-level of detail in comparision GIS maps, and they encourage dynamic, real-time view changes, as well as personalization of the map through annotations. CONCLUSIONS A number of different community applications enabled by interactive maps have been presented. All of these have focused on new collaborative activities that are available with interactive maps. These applications indicate the numerous opportunities for interactive maps to support collaboration. REFERENCES 1. Carroll, J.M., Rosson, M.B., Isenhour, P.L., Ganoe, C.H., Dunlap, D.R., Fogarty, J., Schafer, W.A. and Van Metre, C.A. (2001). Designing Our Town: MOOsburg. Accepted for publication in International Journal of Human-Computer Studies. 2.

Greenberg S. and Roseman, M. (1998). Using a Room Metaphor to Ease Transitions in Groupware. Research report 98/611/02, Department of Computer Science, University of Calgary, Calgary, Alberta, Canada, January.

3.

Mynatt, E.D., Adler, A., Ito, M., and O’Day, V. (1997). Design for Network Communities. In Conference Proceedings of CHI’97: Human Factors in Computing Systems (pp. 296-304).

4.

Schuler, D. (1994). Community Networks: Building a New Participatory Medium. Communications of the ACM, 37(1), 39-51.

X. Appendix C – Technical Report. A Classification of Interactive Map Software Wendy Schafer INTRODUCTION People use paper maps for a number of different activities. Atlases and other maps are often used when traveling in an unfamiliar area, such as driving in a city or hiking in a wilderness area. An adventure novel may include a map so that the reader has a reference for the characters' travels, and most shopping malls have a map to list stores and indicate their locations. Considering all of the different uses for maps, numerous map software programs have been written to simplify and automate some of the many map tasks. Route planning software is frequently used when going on a trip and looking at a computergenerated weather map on the Web is a common task. Our interest is to design and evaluate map applications to support multiple users collaborating on a spatial task. For example, MOOsburg is an online, collaborative environment based on the town of Blacksburg, Virginia (Carroll et al., 2001). Community members use this environment to communicate with their neighbors. The application allows them to collaborate synchronously or asynchronously through chat, a whiteboard, and other objects. The structure of the environment consists of distinct places that parallel real locations in the Blacksburg. Visiting one place at time, users can interact with the different objects and contribute to the different discussions associated with each location. A map widget supports these collaborative activities as it provides a way to navigate and develop the environment. It displays a map of the town’s roads and buildings and marks the "visitable" locations. Users can click on the existing locations or add new locations for other users to visit. The map widget also enables users to find one another as it incorporates awareness information, such as indicating where the current users are located. Yet, this map widget in MOOsburg can be improved and there are other examples where map software can support a collaborative task. Different interaction techniques could make the map easier to use or enable the software to support different tasks. Currently, users just navigate and click on the place they would like to visit. Additional interactions might allow users measure distances using the map or store personal information on the map display. New features could also enable different forms of collaboration. For example, the map itself could become an object for users to collaborate with, allowing users to share ideas on a map display. We can easily differentiate between different map applications, such as the MOOsburg map widget and a map program on the web, based on the tasks they are intended for, but this does not show how the software programs are fundamentally different. A classification of map software will allow us to make these distinctions. It will enable us to compare software that may seem very similar and contrast software that appears very different. It will also reveal different modifications we could make to the MOOsburg map widget and encourage new types of map software. Map software programs can be different in a number of ways. For example, we could differentiate between the applications’ hardware and software requirements or the usability of the user interfaces. Our classification focuses on the interactions the map supports. Supported interactions define what the user can do with the map display. This, in turn, enables different map-based tasks. For example, a map application that supports multiple interaction techniques could possibly be used for a variety of tasks.

Creating a classification of interactivity not only gives us a way to categorize existing map programs in terms of their interactions, but it provides a means to explore new interaction combinations that can reveal novel map-based tasks and unique applications that have yet to be developed (Ellis et al., 1991). This paper reports on the classification of interactivity produced as a result of surveying forty map applications. An explanation of the classification follows along with specific examples from the survey. Unique combinations are also presented, each of which warrant further investigation. CLASSIFICATION OF INTERACTIVITY Forty interactive map applications were surveyed in order to create a classification of map software (see Appendix A). This survey consulted web-based map programs, map software developed for various research projects, and off-the-shelf software products. The map applications were found from a combination of Web searches, personal references, and a research literature search. The set of map programs surveyed provide a good cross-section of interactive map software as the user tasks range from sharing travel information to viewing hydrographic data to creating a military plan. A description of the possible user interactions was recorded for each application, which was then generalized into a classification of interactivity. The classification consists of three dimensions: navigation techniques, support for collaboration, and data sources available (see Figure 1). Navigation techniques refer to the way a user can change the viewpoint of the map, such as through zooming and panning. Support for collaboration refers to an application's ability to allow multiple people to communicate ideas. For example, the map software may allow users to chat in real-time or write comments on the map for later viewers. Data source availability refers to the options available for displaying different data on the map, such as displaying different layers.

Navigation Techniques Navigation with an interactive map program can occur in two different ways. The software can use multiple maps, each with one view, or it can use one map with multiple views. In the first case, the user has little control over how the map is navigated. One chooses between distinct locations that cause a new map to be displayed. A map on a web page that uses an imagemap configuration is a good example. The map contains hotspots that link you to another web page that can contain another map image. When navigation does not occur through multiple maps, users navigate one map with multiple views using some form of zooming and panning. Maps using this type of navigation are classified using two distinct aspects: panning support and zooming support. Panning refers to the way a user can change the area displayed with the map. For example, one might shift the map to the left in order to view something that was hidden off to the right. Zooming refers to the different magnification levels that can be viewed. For example, one might want to zoom out in order to view the entire map area on their display. Within both the panning and zooming components a map application can exist at one of four levels. At the first level, the map software does not allow this type of navigation (panning or zooming). The next three levels incorporate the different ways for supporting navigation: through discrete interactions, continuous interactions, or both.

Multiple Maps

OR

Zooming Support

Navigation Techniques

Discrete and continuous interactions Continuous interactions Discrete interactions No Support No support

Discrete Interactions

Continuous Interactions

Discrete and continuous interactions

Support for Collaboration

Asynchronous Support

Panning Support External resource and map interface support Map interface support External resource support No Support

No support

External resource support

Map interface support

External resource and map interface support

Synchronous Support

Data Source Availability

One data source

Multiple data sources viewed separately

Noninteractive

Layers

Figure 1. A diagram of the classification of interactivity for map software.

When navigation occurs through discrete interactions, using either panning or zooming, the map display refreshes after each interaction. So if a user decides they want to pan or zoom, he/she would interact with the interface in order to issue the command and then wait for the map to refresh. A common implementation is to use buttons in the interface that relate to the viewpoint changes. For example, a user may click on a magnifying glass to increase the zoom or click on a compass to change the area displayed.

Discrete zooming can also occur through drawing a box around an area to zoom into. After drawing the box, the user waits for the display to refresh. Another discrete panning design uses clicking and dragging. Users click on the map and drag it to pan, but the newly exposed areas are not rendered while dragging. Again, once the interaction with the interface is complete, the map display is refreshed and the exposed areas are rendered. On the other hand when navigation occurs through continuous interactions, using either panning or zooming, the display provides continuous feedback to the user. So if a user changes the viewpoint using continuous navigation, the display changes while he/she is performing the interaction. This is commonly implemented with some form of dragging. For example, a user might change the zoom level by clicking and dragging on the knob on a slider widget. Each drag movement will cause the map display to update with every zoom level encountered. Similarly, continuously panning might be performed by clicking and dragging the map, where the display is continuously rendered with each movement. Map applications that use both discrete and continuous interactions for either panning or zooming incorporate multiple techniques for navigation. For example, a map program might support zooming using a discrete interaction, such as clicking on zoom levels, and using a continuous interaction, such as dragging a slider through the zoom levels. Using both forms of navigation, the map enables many different interactions to take place. Typically a map application uses either multiple maps or panning and zooming, although both could be employed. For example, a map application might use multiple maps, each of which could be navigated with multiple views. In this case, navigation primarily occurs through panning and zooming and such a map would be classified in terms of its panning and zooming support. Taking the multiple maps possibility into consideration along with the four levels of zooming and panning support, there are seventeen different possible combinations for navigation. Support for Collaboration The support for collaboration dimension has a similar classification scheme to that of navigation techniques. It consists of two distinct components, asynchronous support and synchronous support, each of which contain four possible levels of interactivity (Ellis et al., 1991). A map application provides asynchronous support if the software allows multiple users to exchange information over time. For example, one user might leave a comment for their colleagues to see the next time they use the map software. Users involved in asynchronous communication do not have to be at different locations, but this is usually the case as many map programs are accessed from a personal workspace rather than a common area. Synchronous support is provided when the software enables multiple users to work together at the same time. Two users holding an online discussion about a map is a good example. Again, users do not have to be at different locations, but this is often the case. Within the two components of asynchronous and synchronous support, a map application can exist at one of four levels. At the most basic level, there is no support for this type of collaboration (synchronous or asynchronous). The next three levels incorporate the different means for supporting collaboration: collaboration using an external resource, collaboration using the map interface, and collaboration using both an external resource and the map interface. If multiple users communicate using a tool other than the map interface than an external resource is supporting collaboration. For example, many web-based map applications have an email feature that sends a URL for a particular map display. Recipients of the email message might simply view the map or they might change the display through navigation, adding content, etc. and then send another message. In

this case, collaboration is supported through the external resource of email. Many other examples occur when the map software is embedded in a larger application that includes collaboration tools. For instance, a synchronous chat feature is a common collaboration tool and a good example of an external resource. The map interface supports collaboration if multiple users can work directly with the map display. When this is the case, the map software itself includes a means for users to communicate and share ideas. For example, a map program might have a collaborative editing tool so that multiple users can contribute to the data set and its representation. Another good example is a map application that includes a shared navigation feature. This would allow multiple users in different locations to see the map from the same viewpoint and watch the same navigation transitions. Map applications that support collaboration both through an external resource and through the map interface incorporate multiple techniques for collaboration. For example, a map application might support collaboration using both a synchronous external resource, such as chat, and a synchronous map feature, such as shared navigation. Using both forms of collaboration support, the map software enables many different interactions to take place. Taking the two components and four levels of collaboration into consideration, there are sixteen different possible combinations. Data Sources Available The third dimension of the classification characterizes the available data sources for the map software. In particular, this dimension looks at whether or not the user's actions can have any effect on the data displayed and if so, how much control is given to the user. For example, a map application may only have one data set and so none of the user's actions can result in different data sets to be used. On the other hand, the map software may include numerous data sets and require that the user specify every data source that is to be displayed. Under the data source dimension, there are four different levels at which a map application can exist. The first level is the most basic and does not give the user any data options. The map uses only one data source and always displays that same data source. A map image on a web page is a good example. It displays one data source and do not provide any data options. At the next level, a map application uses multiple data sources, but can only display one at a time. For example, a display with a state map could allow users to see different county statistics such as population and density of people. The third level is termed non-interactive data sources. In this case, the user does not have any data options, but the software incorporates different data sets and has an underlying algorithm for their use. This is commonly viewed when zooming with map software, as more data and details are often displayed when the magnification increases. At the last level, the user has total control over the data sources used as he/she chooses which data sets to display. This is commonly referred to as layers and is widely used in Geographic Information Systems (GIS). Survey Results By looking at the survey results, we can provide some validity to the classification. Examining the three dimensions individually, we see that the forty map applications were dispersed across the different combinations (see Figure 2, 3, and 4). Out of the possible seventeen combinations for the navigation dimension, the survey examined seven different combinations. Discrete panning and discrete zooming was the most common navigation technique observed, as many web-based applications were examined. All web browsers support a discrete model of interaction, but most require extra software to support continuous map interactions, resulting in many web-based maps that use discrete navigation.

Similarly, the map survey identified maps in seven of the sixteen types of collaboration support. Many of the map applications surveyed did not support collaboration, but again, many applications were located on the Web and required extra software in order to support collaboration. For example, posting a website that uses multiple maps for navigation is much less complicated than a web-based application that supports synchronous collaboration on the map interface. In terms of the data sources available dimension, all four levels were observed. The entire classification has a possible 1,088 different combinations. Our map survey examined twentynine of these combinations. This means that most of the map applications surveyed were different in some way. Some of the maps were classified the same though showing that the classification does provide an abstraction. Navigation Results

Number of Map Applications

25

20

15

10

5

0 no support

multiple maps

discrete discrete zooming and no zooming and panning discrete panning

discrete zooming and continuous panning

continuous zooming and continuous panning

Navigation Dimensions

Figure 2. Map survey results from the navigation dimension of the classification.

discrete zooming, continuous zooming, and continuous panning

Support for Collaboration Results 30

Number of Map Applications

25

20

15

10

5

0 none

asynchronous using an external resource

synchronous using an external resource

asynchronous using map interface

synchronous using map interface

asynchronous and asynchronous using synchronous using an external resource an external resource and synchronous using map inteface

Collaboration Dimensions

Number of Map Applications

Figure 3. Map survey results from the support for collaboration dimension of the classification.

Data Source Availability Results 20 15 10 5 0 only one data single source source display

noninteractive

layers

Data Sources Available

Figure 4. Map survey results from the data source availability dimension of the classification. Other Possibilities While establishing this classification, a handful of other dimensions were considered but not included. These dimensions focused on other interactions that map applications commonly support. For example, many map programs provide customization features, such as changing the color scheme or modifying an icon. Another interesting feature is the ability to add user-defined content to the map. On the one hand, some programs simply allow a user to mark a location with a pinpoint, while others include a range of drawing and annotation tools. A feature that is less frequently implemented is a transformation tool. Such a tool enables users to apply projections and rotations to a map.

Each of these interactions was considered to be part of the classification, but they did not provide very interesting distinctions. These interactions simply modify the appearance of the map as opposed to the interactions included in the classification. The dimensions examine more complex issues such as how users can collaborate with one another, how users can alter the map's viewpoint, and how users can modify the data sources displayed. For example, if we compare two maps that have the same navigation techniques, the same collaboration support, the same data source availability and different combinations of the appearance modifying interactions the combinations produced are not that interesting. In essence, each combination has a different set of neat features and including these combinations in the classification is not advantageous. MAP SOFTWARE EXAMPLES Having described the classification of interactivity, we will now go through a few examples. Each of these examples comes from the survey that was used to create the classification. They are some of the more interesting map applications examined and each exhibits different aspects of the classification. Town of Blacksburg GIS The Town of Blacksburg, Virginia includes a Web-based GIS system on their website. This map application allows town's citizens and visitors to access local GIS data. It provides information regarding the town's street system, property boundary lines, bus stop locations, zoning, and public locations, such as parks and libraries. The map is primarily used for reference and most tasks involve using the map to find a specific piece of local information (see Figure 5). Looking at the software in terms of the classification, the map supports a fairly common set of interactions. The most prominent feature is the maps use of layers. Users can turn layers on and off through a series of checkboxes and an "Apply" button, where clicking on the button causes the map to update. Layer choices include area-based information such as zoning, line-based information such as roads, and point-based information such as libraries. This use of layers is particularly unique in that more layer choices appear as the zoom magnification increases. In terms of navigation, panning occurs by clicking on the triangles on the edges of the map or by setting the next click to be the new map center. Zooming occurs in three increments, 2x, 4x and 8x, and happens through a drop-down menu and a click on the map. Alternatively, clicking on the "Location Map" and the zoom scale will change the area displayed and the magnification, respectively. This map suffers from a common web problem in that all interactions require a message to be sent to the web server so that the screen can refresh. The map supports a couple of other interactions such as searching for an address, listing public places, and identifying different aspects of the display, but it does not support any collaborative activities. If two people want to look at this program together they need to be physically located at same computer at the same time. The software does not include a way for multiple people to know that they are using the application at the same time, nor does it support multiple collaborators using it over a period of time.

In terms of the classification of interactivity, this map application uses a common combination. It has no support for collaboration, navigation occurs through discrete zooming and discrete panning, and data sources are accessible through layers.

Figure 5. A screenshot from the Town of Blacksburg’s WebGIS, a map application that uses discrete zooming and panning and a layers data source. Triscape Map Explorer Triscape is a company founded in 1997 whose mission is to develop virtual reality software that solves real world problems. Their first product, Map Explorer, is a unique software application involving road maps, web pages, and user-created annotations. One focus of the map software is to allow companies to publish web pages and paintings spatially, in order to sell a product or service. Customers enjoy a unique experience as they navigate this spatial layout and learn about the company. The other goal is provide an alternative to tradition on-line map programs. As a result, the map software is designed for three basic tasks: creating a company space, customizing an on-line map, and exploring these creations (see Figure 6). The map software supports a range of interactions including continuous zooming, continuous panning, annotations, virtual links, and displaying live web pages. Using the standard navigation mode, the compass widget is used to continuously zoom and pan the map interface. Holding the mouse down on the arrows and the plus and minus symbols causes the map to pan and zoom, respectively. Alternatively, clicking and dragging the mouse on the map allows user to zoom and pan simultaneously. A left click corresponds to zooming out, a right click is mapped to zooming in, and the middle button provides free form continuous panning.

The map also supports navigation through virtual links and common Web browser interactions such as "Back" and "Forward". If a user clicks on a virtual link the map automatically navigates to a new location using continuous interactions. Using the "Back" and "Forward" buttons, one can retrace their steps. Changing the magnification affects the amount of detail displayed on the map. Zooming in results in data being added and zooming out results in data being removed. The application uses multiple levels of detail with each building upon the last. They include major highways and major cities, important roads, and minor roads. Labels are also added at different levels of magnification. This map software has a different approach than traditional on-line map applications such as Mapquest and MapBlast. It provides a range of annotation features so that users can customize their maps. One can add a wide highlighting line, a local road, a pond, a link to a live web page, a user-specified line or area, etc. Once the map is customized, the map can be saved on the web and referenced through a static URL for a year. This allows a user to have other people review their annotations and add new ones. In terms of the classification of interactivity, this map application uses a unique combination. It has asynchronous support for collaboration using the map, navigation occurs through continuous zooming and continuous panning, and data sources are not accessible but rather used non-interactively.

Figure 6. A screenshot from Triscape’s Map Explorer. This map application supports continuous zooming and panning, asynchronous collaboration using the map interface, and non-interactive data sources. VirtualTourist VirtualTourist is website for travelers to share information. The website includes a range of activities to participate in as well as various travel tools such as currency converters, time zone maps, and current travel warnings. Over 146,000 people from 223 countries use this site to meet other travelers and share experiences. People share pictures, travel tips, travel stories, and reviews of destinations. The web site is organized so that each continent, country or territory, state, and city has a web page. On these pages, one

can read user-posted comments relating to general tips, restaurants, accommodations, must see activities, nightlife, tourist traps, transportation, and so on. These pages also list registered users who have created their own page dedicated to the location. Typically, registered users create their own set of web pages that follow the same structure as the general use pages. These pages allow them to add travel experiences and pictures and give them a homepage for others to learn about their interests. VirtualTourist uses a map interface for navigating their web pages. A map is included at the top of every web page so that users can choose to navigate to a subsection or read about the current area. For example, on the web page for the United States, a map of the US with state boundaries and state labels is displayed and each of the states are linked to the web page about that state. Using the maps, a user can easily drill down to the location they are interested in. Reversing the process is less intuitive as one has to either use the back button or recognize the other navigation aids on the WebPages (see Figure 7). The maps used on the general use WebPages always look the same and there are no options to modify them. Personal web pages are marked with a color-coding based on the information provided by the user. Areas and places where the user claims to have been are colored with red, places the user plans to go are colored with green, and places where the user has been before and plans to go again are colored purple. This makes for a colorful map, but there is no real control over what data is displayed. Using the VirtualTourist, users can also participate in one of the many discussion forum or chat online in real-time. The discussion forum organization parallels that of the web pages. Each continent, country or territory, state, and city has a separate discussion associated with it. Initial postings are typically questions, which are then followed by a series of replies. The online chat contains about 40 distinct rooms that relate to different travel topics, such as California, German cheese, and cafes. Other interesting activities include sending an electronic postcard with a picture from almost anywhere in the world and searching for people with similar interests or with specific attributes in their profile. In terms of the classification of interactivity, this map application uses another unique combination. It has synchronous and asynchronous support for collaboration using an external resource, navigation occurs through multiple maps, and only one data source is used.

Figure 7. A screenshot from the VirtualTourist website. This map application uses multiple maps for navigation, synchronous and asynchronous collaboration through an external resource, and only one data source. EXPLORING THE CLASSIFICATION A classification of interactivity enables us to do more than just classify existing maps. It also encourages the exploration of different interaction combinations. Given our multiple component classification, we can explore different combinations within each dimension as well as between dimensions. This section focuses on combinations within the dimensions of navigation techniques and support for collaboration. It is interesting to look at the advantages and disadvantages of each unique feature within a dimension. This allows us to attribute the different features to different map characteristics, which can be used to analyze new interaction designs. Some combinations of features are complementary and provide more power or flexibility than simply the sum of the two features, while others are simply additive. Navigation Techniques The dimension of navigation techniques has seventeen different possible combinations using multiple maps and four levels of panning and zooming support (none, discrete interactions, continuous interactions, and both discrete and continuous interactions). We examine only some of these combinations.

The use of multiple maps is a good approach when there are a small number of views to see. One possible scenario is a map application that displays the approximate locations of a collection of landmarks. The map display might include major roads and a clickable icon for each landmark, where clicking on an icon would bring up a new map focused on the landmark and the nearby roads. On the other hand, the use of multiple maps has the distinct disadvantage that navigation is very limited. A map application that only uses zooming or panning has restricted navigation. With just zooming the map is limited in the area that can be viewed and with just panning the map is limited in the details that can be viewed. This is the case for both discrete and continuous panning and zooming. Many systems that use discrete interactions suffer from slow refresh rates. For example, using a map program on a webpage often involves waiting for a map image to refresh. These slow refresh rates can be frustrating to a user. On the other hand, discrete interactions enable precision. For example, discrete panning might allow a user to shift the map in respects to a distance measurement or discrete zooming might allow a user to magnify the map by exactly 2x. Discrete zooming is unique in that different implementations can have different consequences. For example, if the user can indicate a rectangle area to zoom into, then the interaction supports a form of panning as well as zooming. This implementation also gives the user greater control over zooming as it allows one to specify an area to magnify as opposed to a design that zooms in towards a particular point. In comparison, continuous interactions have an advantage in that users get continuous feedback about the action they are performing. This feedback supports a closed-loop interaction, it enables them to stay oriented as well as evaluate their decision and potentially alter the interaction. The feedback also helps a user to quickly learn how the map responds. For example, using a continuous zoom and magnifying the map as much as possible, one can easily see how the magnification and the associated levels of detail can change. Another advantage of continuous interactions is their use of free-form movement. A user can use any number of zoom levels using continuous zooming as opposed to the fixed number available with discrete zooming. On the other hand, continuous interactions may be difficult to control. It is possible for a user to have trouble with the movement associated with continuous interaction resulting in the map changing undesirably. While exploring the different combinations of discrete and continuous panning and zooming, two combinations stood out as being particularly interesting. First, a map application that implements both discrete zooming and continuous zooming can be useful. Both approaches have distinct advantages, which can be combined to produce a more versatile map application. As mentioned earlier, discrete zooming where the user specifies a rectangle area gives the user more control over what is magnified and continuous zooming offers a better feedback technique. When used in the same application, users can better specify their navigation with two unique techniques, although the interface is more complex. The combination of discrete panning and continuous zooming is also interesting to consider. This combination could be helpful in tasks that are more detailed-oriented. One scenario might be where a user wants to learn about the area around a particular place and get a perspective on that place's location within a larger context. Such a task involves minimal panning and extensive zooming. Collaboration The dimension of collaboration has sixteen different possible combinations resulting from the four levels of asynchronous and synchronous support (none, support for collaboration using an external resource,

support for collaboration using the map interface, and both support for collaboration using an external resource and using the map interface). We examine only a few of these combinations. A map application that only supports asynchronous or synchronous collaboration can be restrictive. With just asynchronous collaboration the map software does not allow for chance encounters. For example, if two people happen to be using the map software at the same time, the lack of support for synchronous collaboration could limit their communication and even prevent them from identifying their simultaneous use. On the other hand, if a map application simply supports synchronous collaboration, the software is limited in its use. In order for one user to share their ideas, another user has to be using the application at the exact same time. If this is not the case, then the user is forced to remember it until someone else is using the application or share the idea using a different asynchronous tool. The second alternative may seem like a reasonable solution, but it can be difficult to specify a spatial idea without using the map application that generated the idea. When the only support for collaboration is directly through the map interface, all communication must be conveyed on the map itself. This might be sufficient for some map intensive activities, but external resources could also be helpful. For example, a user might want to leave a message that does not relate to the spatial display. Collaboration support using the map interface forces the user to post the message on the map, whereas an external resource would allow the message to part of a different information representation. Placing all communication on the map interface also can make the map cluttered and difficult to read. In comparison, when collaboration is only possible through external resources, no communication can be conveyed on the map. This could provide sufficient support for activities that are not map intensive, such as discussing general trends in the display, but communicating on the map might also be useful. For example, a user may want to illustrate their idea by physically pointing out an area on the map or annotating on the map for others to see. Collaboration using just an external resource does not allow these interactions to occur. One combination that is interesting to consider is support for synchronous collaboration using both an external resource and the map interface. This provides good support for many synchronous activities where people can collaborate using a range of tools. But using multiple tools could be overwhelming, as one has to pay attention to many activities during a session. NOVEL COMBINATIONS Based on our survey, it seems that maps applications that do not support some form of collaboration are the majority. This makes it interesting to explore the classification with a focus on collaboration support. Using different combinations of support for collaboration along with different instances of navigation techniques and data sources available, we examined many points in the design space. Most of the combinations were not that different from an existing map application, but some of instances were novel. One of the novel combinations identified uses asynchronous and synchronous collaboration support through an external resource and a layers data source. This is unique because most map applications that feature layers offer no support for collaboration, and those that do often focus on asynchronous collaboration using the map interface. In contrast, this particular combination allows multiple users to share ideas and explore different combinations of data sources. It enables users to work independently with the map and then contribute their ideas either in real-time or over time. One scenario that uses this combination occurs when multiple people in different locations are planning a trip together. The users would carry on a discussion about the different places they want to visit, when

and where they could meet up, the amount of time they want to spend at each location, different route possibilities, etc. The map would aid their decisions by displaying different data sources such as transportation terminals, major hotel locations, historical sites, interstate highways, and so on. The users would collaborate when everyone was using the application, when a subset was using the application, and when only one member was contributing ideas. This combination has the disadvantage that it can be considered too similar to the VirtualTourist example to be truly unique. With the VirtualTourist, users discuss travel destinations by creating web pages with their comments and by using an online chat tool. The map displays an image of the place being discussed and it is navigated through multiple maps. With this combination the activities and interactions are similar, as users would use the map application to discuss different views of the data. Examining a slightly different combination, the similarity is not present. Take for example a map application that supports asynchronous and synchronous collaboration using an external resource, synchronous collaboration using the map interface, and a layers data source. This is identical to the previous combination only now we have added support for synchronous collaboration using the map interface. This allows users not only to discuss the different views of the map but also to collaborate on the map display itself. For example, the software might allow users to share annotations on the map or enable users in different locations to view the same layers but navigate the map independently. Adding this extra component opens up a range of possible interactions and makes the map application different from that of the VirtualTourist. This combination could be particularly useful for the field of GIS. For example, one scenario involves a collection of users analyzing the best place to put an interstate toll collection barrier. Layers would allow environmentalists, town's people, and commercial traffic companies to express their individual views as well as see each other's points (Gallo, 1996). Users could discuss the matter at the same time adding annotations to the map in order to express their points. Alternatively, users could view the data on their own at different times and become abreast of the different positions. Along the same lines, extending the last combination to include asynchronous collaboration through the map interface adds even more interaction opportunities. With this combination, multiple users can communicate spatial ideas in real-time as well as over a period. This means that two users do not have to be using the application at the same time in order share ideas directly related to the map display. This combination seems particularly useful in a community application, such as MOOsburg. Unlike a traditional work setting, people contribute to community-based activities at all hours. For example, someone may log on to MOOsburg late at night to add his or her comments to a town discussion forum. This type of environment requires both synchronous and asynchronous collaboration support. A map combination that includes asynchronous and synchronous collaboration through the map interface and external resources provides this. Also, the use of both the map interface and an external resource allows for a variety of discussions to occur. Looking at these last two combinations closer, we can generalize our results in that all maps that include synchronous collaboration support using the map interface and a layers data source are novel. Most map applications that feature layers do not support collaboration and those that do offer asynchronous collaboration through the map interface. Our map survey included two examples of synchronous collaboration using the map. Both of them used layers and yet both of them can be further qualified. The Collaborative Map Annotator uses an interactive map on a large screen display for same place collaboration. Users can add and remove different terrain data, route maps, surveillance and intelligence information, and archived GIS data. This example is interesting because it allows users collaborate in the same location, but if we explore the use of remote collaboration we find a novel combination.

Similarly, MapsOnUs is a web-based map program that allows people to draw maps, plan routes, and search yellow pages. Two users can synchronously use this map by clicking on a link that makes the two views identical. In order to stay abreast of the other person's interactions, the link must be clicked repeatedly. This system does not support real-time synchronous collaboration, which would be desirable when working with a map. CONCLUSION This paper has described a classification of map software based on interactivity. It has provided three examples as well as discussed the classification. Also, we have explored some of the unique combinations that were revealed by the classification. Creating such a classification encourages us to analyze map programs in terms of their fundamental characteristics. It causes us to think beyond the software's users, tasks, and work environment. This paper presented one way to analyze map programs. We looked at the software in respects to its navigation techniques, support for collaboration, and data source availability. This analysis technique serves as an example and encourages other means for analysis to be developed. In particular, this classification enabled us to explore new interaction possibilities. It allowed us to go beyond what has been implemented and consider the map activities that are possible through different interaction combinations. These activities are still ideas, but they can be evaluated and supported in new map software applications.

APPENDIX This is a listing of the forty map software applications surveyed during the creation of the classification of interactivity. Name

Location

Navigation Technique

Support for Collaboration

Data Sources Available

ActiveMap Andersen continuous synchronous using one data source (McCarthy et al., Consulting panning and an external 1999) discrete zooming resource The map serves as an at-a-glance awareness tool that shows where people are within an office space. ArcData Online

www.esri.com/dat a/online/

discrete zooming no support non-interactive and discrete panning A web application that allows users to create maps using data for sale at ArcData Online ArcView GIS 3.2

ESRI product

discrete zooming no support layers and discrete panning A GIS produce that provides data visualization, querying, analysis, editing, and integration capabilities. Autodesk www.mapguide.co discrete zooming no support layers MapGuide m and discrete LiteView panning Extension An on-line application that serves maps without having users download and install a special viewer. Autodesk MapGuide Viewer

www.mapguide.co m

discrete zooming asynchronous and discrete using the map panning interface Software that delivers custom-specific maps and design data easily across the Web.

layers

Bay Area Rapid wwwno support no support multiple data Transit (BART) itg.lbl.gov/vbart/ho sources viewed Schedule mepage.html separately Animation A simulation that provides an infrastructure visualization of San Francisco’s transit system. City of Tucson ArcIMS demo

www.ci.tucson.az. us/ed/ed.htm

discrete zooming no support layers and discrete panning An on-line application to find and compare vacant commercial properties in Tucson, Arizona. Collaborative Map www.rl.af.mil/tech no support synchronous using Annotator /programs/ADII/ad the map interface (Jedrysik, 2000) ii_cma.html A demonstration application for the Interactive Data Wall research project.

layers

EPIC Web Browser

www.epic.noaa.go v/epic/ewb

discrete zooming

no support

multiple data sources viewed separately

An on-line tool to provide access to EPIC hydrographic data sets. EtakGuide Map

www.etakguide.co m

discrete zooming no support and discrete panning A demonstration of geocoding, a procedure to find and display any address on a map.

non-interactive

Geophysical Fluid Dynamics Lab Data

www.cdc.noaa.gov no support no support multiple data /cgisources viewed bin/DataMenus.pl? separately dataset=gfdl An on-line tool to visualize global climate variable data from the Geophysical Fluid Dynamics Lab. GeoMedia Web Map

www.intergraph.co m/gis/gmwm

discrete zooming no support and discrete panning Commercial software produce that places GIS data on the Web.

layers

GRASSLinks

www.regis.berkele discrete zooming no support layers y.edu/gldev/regis.h and discrete tml panning A Web interface to a GIS system that facilitates data sharing between environmental planning agencies, public action groups, citizens, and private entities. Hourly/Daily Rain Data

precip.fsl.noaa.gov discrete zooming no support layers /hourly_precip.htm and continuous l panning An on-line tool to display daily and hourly precipitation totals from various locations all over the US. Houston RealTime Traffic Map

traffic.tamu.edu/in cmap/incmap.asp

no support

no support

multiple data sources viewed separately

A web-based application to convey real-time traffic conditions in Houston, Texas. Interactive icemaps.des.ucdav discrete zooming California is.edu/icemaps2/IC and discrete Environmental EMapInit.html panning Management, Assessment, and Planning System (ICEMAPS) A Web interface to an environmental, California GIS.

no support

layers

Interactive WebMapping Tool

no support

layers

atlas.geo.cornell.e du/ima.html

discrete zooming and discrete panning Cornell’s Geoscience Information System tool on the Web.

Image Web Server

www.earthetc.com /ecwcounty/ecw_c ounty_frame.asp

discrete zooming, no support one data source continuous zooming, and continuous panning A demonstration of Image Web Server technology – county assessor data is provided on-line. JShape Software

www.jshape.com/i ndex0.html

discrete zooming no support layers and discrete panning A Java software package that supports fast and easy creation of interactive, web-based, GIS applications. Macromedia www.stmaartenst discrete zooming Shockwave martin.com/freeha and discrete FreeHand maps nd panning This software displays a FreeHand file on the web. MAPBLAST!

no support

one data source

www.mapblast.co m

discrete zooming asynchronous layers and discrete using an external panning resource An on-line application to drawn maps of anywhere in the US and provide routing information. Map-It

crust.er.usgs.gov:8 no support no support one data source 0/mapit A demonstration of the free, open source software package GMT. This on-line tool draws a map of usersupplied latitude and longitude pairs. MapQuest

www.mapquest.co m

discrete zooming asynchronous layers and discrete using an external panning resource An on-line application to drawn maps of anywhere in the US and provide routing information. Maps.com

www.maps.com/le arn

Maps On Us

www.mapsonus.co m

discrete zooming and discrete panning

MOOsburg map widget

moosburg.cs.vt.ed u

continuous zooming and continuous panning

discrete zooming no support and continuous panning Education resources to make learning fun. Maps are animated and interactive.

layers

asynchronous layers using an external resource and synchronous using the map interface An on-line application to drawn maps of cities and yellow page listings in the US and to provide routing information. asynchronous and synchronous using an external resource

non-interactive

panning resource A navigation tool, awareness tool, and end-user developer tool for MOOsburg. MyWay.com’s javamap

www.zip2.com

discrete zooming and discrete panning An on-line application to drawn maps of anywhere in the US.

asynchronous using an external resource

non-interactive

NOAA Real-time www.pmel.noaa.g no support no support multiple data TAO Buoy Data ov/togasources viewed Display tao/realtime.html separately An on-line tool to display real-time data from ocean buoys in order to analyze El Niño and La Niña. Pacific Century lps.pcgsys.com/w discrete zooming no support Systems Limited’s ml/elpas/html and discrete GIS Map panning A password-protected website to track co-workers’ locations at PCS.

layers

Portland State www.ncn.com/~ril no support no support University Child g/Mapper/ore.htm Welfare Partnership Thematic Mapping Program A web-based tool that provides child welfare data for the state of Oregon.

multiple data sources viewed separately

Puget Sound Traffic Conditions

www.wsdot.wa.go multiple maps no support one data source v/PugetSoundTraff ic A webpage that displays the current traffic conditions for the Puget Sound area in Washington state. Race Track chasinracin.net/tra multiple maps Locator ck-locator A website to locate race tracks around the US.

no support

non-interactive

Ramonamap (Bartlett, 1994)

asynchronous using the map interface

one data source

no support

layers

Western Research Lab

no support

An application to share office information spatially. Rand McNally Rand McNally discrete zooming TripMaker Deluxe and discrete 1999 Edition panning A commercial software product to plan road travel.

Town of www.webgis.net/b discrete zooming no support layers Blacksburg lacksburg and discrete WebGIS panning An on-line tool that provides information regarding the town’s street system, properties, parks, bus stop locations, zoning, etc.

Taiwan Map

peacock.tnjc.edu.t w/ADD/maps/taiw anmap.html A primitive interactive map website. Triscape Map Explorer

multiple maps

no support

continuous asynchronous zooming and using the map continuous interface panning An on-line application that provides an alternative approach to on-line mapping.

non-interactive

Virtual Slaithwaite www.ccg.leeeds.ac discrete zooming asynchronous (Evans et al., .uk/slaithwaite and discrete using the map 1999) panning interface A demonstration of a local, environmental decision-making, public participation tool.

one data source

Virtual Tourist

www.triscape.com

non-interactive

www.virtualtourist .com

multiple maps

asynchronous and only one data synchronous using source an external resource A web-based application that allows people from all over the world to share travel plans.

REFERENCES Bartlett, Joel F. “Ramonamap: An Example of Graphical Groupware.” In proceedings of the ACM Symposium on User Interface Software and Technology (1994): 83 - 84. Card, Stuart K., Mackinlay, Jock D. and George G. Robertson. “The Design Space of Input Devices.” In proceedings of Computer-Human Interaction (1990): 117 - 124. Carroll, John.M., Rosson, Mary.Beth, Isenhour, Philip.L., Van Metre, Chirstina.A., Schafer, Wendy A. and Craig H. Ganoe. “MOOsburg: Multi-user Domain Support for a Community Network.” Internet Research 11:1 (2001): 65-73. Ellis, Clarence A., Gibbs, Simon J.and Gail Rein. “Groupware: Some Issues and Experiences.” Communications of the ACM 34: 1 (January 1991): 39 – 58. Evans, Andrew, Kingston, Richard, and Ian Turton. “Web-based GIS to Enhance Public Democratic Involvement.” Geocomp99 Conference Proceedings (1999). Gallo, Sharon, K. “GIS Within Electronic Meetings: The Effective Use of Spatial Data in a Joint Intellectual Effort.” In proceedings of the ESRI International User Conference (1996). Jedrysik, Peter A., Moore, Jason A., Stedman, Terrance A., and Richard H. Sweed. “Interactive Displays for Command and Control.” IEEE Aerospace Conference Proceedings 2 (2000): 341 -351. McCarthy, Joseph, F. and Eric S. Meidel. "ActiveMap: A Visualization Tool for Location Awareness to Support Informal Interactions." In proceedings of Handheld and Ubiquitous Computing (1999): 158170.