The final outcome of my thesis questions, research, and design explorations focused on four design prototypes, each intended to support a different stage in a user's decision-making process for finding a neighborhood that meets their unique combination of preferences and needs.
- Prototype 1: Articulating and weighting user-defined preferences
- Prototype 2: Discovering and saving converging preferences
- Prototype 3: Exploring neighborhoods and citizen-generated content
- Prototype 4: Sharing neighborhood maps and on-site documentation
PROTOTYPE 1
Articulating and weighting user-defined preferences
In this first prototype, I designed a process that scaffolds the user’s definition of living preferences. The user is presented with a list of generalized, preference categories, each related to city-living (these include “Food Options,” “Arts and Entertainment,” “Public Safety,” etc.). The user is able to enter customized categories at this time. The context-aware system is then able to prompt the user to choose or add search terms within each category, allowing the user to better articulate what he/she intends to search. A user may add “restaurants,” “bagels and co,” and “Whole Foods” under the category “Food Options.”
This hierarchical structure of generalized preference categories followed by specific search terms is advantageous for two reasons. First, this structure allows the user to think of each preference in both broad and specific terms. Defining specific instances of the category “Food Options” helps the user generate a variety of satisfactory choices, while at the same time recognizing that general access to food is what is of true importance. This might prevent a user from only searching for places to live near restaurants, for example. Second, this structure anticipates the user’s ability to refine his/her search preferences throughout the decision-making process to be either absolute or negotiable, depending on what is displayed on the map.
This process of refining one’s preferences by way of prompts both externalizes and articulates what may otherwise be left to internal, unstructured intuition. This externalization is vital for a user to reflect on and improve upon what is important to him/her with regards to a living environment.
Once multiple preference categories are defined, the system allows the user to weight the importance of each one relative to the others by means of horizontal sliders. Based on the forced-total principle from Thomas Saaty’s (1990) Analytic Hierarchy Process, the increase in weight of one preference results in the decrease in weight of all others. In this way, the system forces the user to prioritize his/her preferences, providing an inherent mechanism for justification.
This hierarchical structure of generalized preference categories followed by specific search terms is advantageous for two reasons. First, this structure allows the user to think of each preference in both broad and specific terms. Defining specific instances of the category “Food Options” helps the user generate a variety of satisfactory choices, while at the same time recognizing that general access to food is what is of true importance. This might prevent a user from only searching for places to live near restaurants, for example. Second, this structure anticipates the user’s ability to refine his/her search preferences throughout the decision-making process to be either absolute or negotiable, depending on what is displayed on the map.
This process of refining one’s preferences by way of prompts both externalizes and articulates what may otherwise be left to internal, unstructured intuition. This externalization is vital for a user to reflect on and improve upon what is important to him/her with regards to a living environment.
Once multiple preference categories are defined, the system allows the user to weight the importance of each one relative to the others by means of horizontal sliders. Based on the forced-total principle from Thomas Saaty’s (1990) Analytic Hierarchy Process, the increase in weight of one preference results in the decrease in weight of all others. In this way, the system forces the user to prioritize his/her preferences, providing an inherent mechanism for justification.
When the user chooses to define his/her living preferences, the map transitions into a dark grid overlay, anticipating the visualization of data.
|
The user is presented with a list of generalized, preference categories, each related to city-living. The user is able to enter customized categories in addition to those provided.
|
The context-aware system is then able to prompt the user to choose or add search terms within each category, allowing the user to better articulate what he/she intends to search. Here, the user adds “Food co-ops” as an additional search term to the Food Options preference category.
|
The system prompts the user to weight the importance of each preference category by means of horizontal sliders.
|
An increase in the weighted importance of one preference automatically decreases the importance of all others.
Decreasing the weighted importance of a preference category weakens its visualization on the map.
|
Increasing the weighted importance of a preference category strengthens its visualization on the map.
|
PROTOTYPE 2
Discovering and saving converging preferences
Once multiple preference categories are defined and assigned a relative weight of importance, the results are visualized on a city-wide view of the map. For each preference, areas of high correspondence to the search terms are given greater visual prominence than areas of low correspondence, creating a unique overlay pattern. Each preference category is assigned a unique color, so the user may distinguish each variable. When preference patterns are layered, the user looks for areas of greater convergence (areas with high concentration of Food Options and Places to Study, for instance). The user is able to isolate areas where these preferences converge, visualized as a neutral gray color. As the user refines his/her search terms, and reweights his/her preferences, the visualized areas of convergence respond accordingly. The user is then able to save these combined preference sets in a “card” format. These cards show a thumbnail view of the weighted preferences and corresponding areas of convergence. The user can then revisit these cards at a later date, as well as share them among other people with whom they are moving.
This exploratory activity allows the user to iteratively investigate the cause and effect of combining different preferences with the changing areas of convergence across the city. The visualization technique is intentionally general at this stage, not to mislead the user towards interpreting a simplified description of place, but to allow him/her to easily see general patterns of preference convergence. The constantly changing areas of focus on the map (due to changes in the user’s preferences) also encourages the user to delay commitment to any one area of the city in particular until multiple preferences are converged.
The similar visual treatment of distinctly different phenomena (food options and bikeable areas, for instance) remains a difficulty. There is a trade-off in visualizing all of one’s preferences in a similar manner (to aid in seeing where they converge) and over-simplifying the differences in phenomena that source these visual patterns.
This exploratory activity allows the user to iteratively investigate the cause and effect of combining different preferences with the changing areas of convergence across the city. The visualization technique is intentionally general at this stage, not to mislead the user towards interpreting a simplified description of place, but to allow him/her to easily see general patterns of preference convergence. The constantly changing areas of focus on the map (due to changes in the user’s preferences) also encourages the user to delay commitment to any one area of the city in particular until multiple preferences are converged.
The similar visual treatment of distinctly different phenomena (food options and bikeable areas, for instance) remains a difficulty. There is a trade-off in visualizing all of one’s preferences in a similar manner (to aid in seeing where they converge) and over-simplifying the differences in phenomena that source these visual patterns.
Each preference category is assigned a unique color, so the user may distinguish each variable. Here, "Food Options" are shown in green...
|
...while "Places to Study" are shown in purple.
|
The user is able to isolate areas where these preferences converge, visualized as a neutral gray color. As the user refines his/her search terms, and reweights his/her preferences, the visualized areas of convergence respond accordingly. Users may layer multiple preference categories at once, and save different combinations of preference sets for later review.
|
Saved preference category sets are displayed in the format of “cards.” Each card displays the relative preference weights, a thumbnail view of the resulting convergence map, and a time-stamp of when it was saved.
|
Users are able to view each other’s saved preference cards. The system allows for combining cards from each user for comparison on the map.
|
When viewing each other’s preference cards on the map, the user can discover areas of convergence, displayed as white. This allows multiple users to see areas of the city that satisfy each of their combined preference sets.
|
PROTOTYPE 3
Exploring neighborhoods and citizen-generated content
Once the users have identified multiple areas of convergence within their preferences, the map allows for a deeper exploration of these areas. By way of a gestural zoom (pinching out with one’s index finger and thumb), the converging patterns resolve from generalized pattern overlays to points of specific location. Data that relates to the user’s search criteria is now visualized in a more detailed manner, on a refined neighborhood map showing streets and local points of interest. Specific locations are given a geographically accurate place on the map, and information about specific locales (names, hours of operation, etc.) can be accessed via tooltips.
In addition to seeing how these locales relate spatially to one another, the system allows the user to view citizen-generated content about the neighborhood in the form of photos and text. At this stage, the user is able to “pull” the top edge of the map down, in essence tilting it into three-dimensional perspective. By then placing a location marker on the map, the system is able to display geo-located photos from social media sites such as Facebook, Flickr, and Instagram that correspond to the marker’s location. Rather than display these images as a grid, the system stitches the photos together into a panoramic collage, based on the direction that the marker is facing on the map. The user may either move or rotate the marker to access other user-generated panoramic photos.
This visualization technique builds on the similar technological capabilities described by former Microsoft engineer Blaise Aguera y Arcas at a 2010 TED conference in which he describes the potential for integrating user-generated content into Bing Maps (Aguera y Arcas, 2010). While Aguera y Arcas demonstrates this capability using one photo at a time culled from Flickr and superimposed on a computer-generated street-view environment, the system I propose here recreates the environment from a multitude of user-generated photos, with the option to filter them based on the time of day they were taken. This feature gives the user opportunity to see the variety of activities (or lack thereof) that occur and are documented in a location, as well as offer new contexts in which each image can be read, due to its juxtaposition with other images in the same location taken by other people. A user might see images of a deserted street by day, but flush with people, cars and lighted signs by night, indicating the range of activity that can happen at given intersection of the city.
In addition to seeing how these locales relate spatially to one another, the system allows the user to view citizen-generated content about the neighborhood in the form of photos and text. At this stage, the user is able to “pull” the top edge of the map down, in essence tilting it into three-dimensional perspective. By then placing a location marker on the map, the system is able to display geo-located photos from social media sites such as Facebook, Flickr, and Instagram that correspond to the marker’s location. Rather than display these images as a grid, the system stitches the photos together into a panoramic collage, based on the direction that the marker is facing on the map. The user may either move or rotate the marker to access other user-generated panoramic photos.
This visualization technique builds on the similar technological capabilities described by former Microsoft engineer Blaise Aguera y Arcas at a 2010 TED conference in which he describes the potential for integrating user-generated content into Bing Maps (Aguera y Arcas, 2010). While Aguera y Arcas demonstrates this capability using one photo at a time culled from Flickr and superimposed on a computer-generated street-view environment, the system I propose here recreates the environment from a multitude of user-generated photos, with the option to filter them based on the time of day they were taken. This feature gives the user opportunity to see the variety of activities (or lack thereof) that occur and are documented in a location, as well as offer new contexts in which each image can be read, due to its juxtaposition with other images in the same location taken by other people. A user might see images of a deserted street by day, but flush with people, cars and lighted signs by night, indicating the range of activity that can happen at given intersection of the city.
Zooming in (pinching out) transitions the general map patterns to show a city roadmap with specific locations relating to the user’s preferences.
|
Specific locations are color-coded to the user’s preference categories, allowing the user to see the density, distance and locations of relevant information.
|
The user is able to access local area reviews collected from neighborhood forums, such as StreetAdvisor. These reviews can be saved for later review.
|
By pulling down on the top edge of the map, the user “tilts” the environment into three-dimensional space.
|
A location marker is placed in the center of the map. Corresponding citizen-generated geo-tagged photos from social media sites are stitched together into a scrollable panoramic image of the area. By tapping on each photo, the user is able to read the original caption. These citizen-generated photos can also be saved for review at a later time.
|
The user is able to alter the images by time of day. Tapping the night icon transitions the images to photos only captured during the evening hours for this specific location.
|
PROTOTYPE 4
Sharing neighborhood maps and on-site documentation
The fourth and final prototype of my thesis addresses the need for multiple people who are moving together to share information. In many cases, an on-site visit is the final step towards making a decision about where to live. Previous research about a city using this system culminates in users saving specific neighborhood locations that are worth investigating in-person. These can also be saved in the form of neighborhood cards that indicate the number of saved citizen-generated photos and reviews. These neighborhood cards can be shared between multiple users who are moving together.
Once in the city, the system allows an on-site user to document his/her experiences through capturing images, video and text captions. Like the user-generated content used in the previous prototype, these images too are geo-tagged and visualized in relation to the user’s neighborhood-scale map. After sequentially documenting areas of the city that appear to offer an ideal lifestyle, the user is able to review his/her experiences in a more spatially-oriented manner than a typical photo grid affords.
Additionally, the system allows for multiple users to view each other’s documentation, as well as collaboratively annotate each other’s photos. This feature is ideal in cases where one roommate is able to visit on-site while the other is only able to view potential areas remotely. Rather than passively viewing content generated by unknown people to build an understanding about a particular area of the city, the roommates here use the map, photos, and captioning capabilities to actively communicate and plan. Through this activity, the users are able to confidently corroborate their previous, remote research of data and citizen-generated content with personal, real-time experiences.
Once in the city, the system allows an on-site user to document his/her experiences through capturing images, video and text captions. Like the user-generated content used in the previous prototype, these images too are geo-tagged and visualized in relation to the user’s neighborhood-scale map. After sequentially documenting areas of the city that appear to offer an ideal lifestyle, the user is able to review his/her experiences in a more spatially-oriented manner than a typical photo grid affords.
Additionally, the system allows for multiple users to view each other’s documentation, as well as collaboratively annotate each other’s photos. This feature is ideal in cases where one roommate is able to visit on-site while the other is only able to view potential areas remotely. Rather than passively viewing content generated by unknown people to build an understanding about a particular area of the city, the roommates here use the map, photos, and captioning capabilities to actively communicate and plan. Through this activity, the users are able to confidently corroborate their previous, remote research of data and citizen-generated content with personal, real-time experiences.
By framing the map in the viewport at the desired scale, the user is able to save preferred areas for later.
|
Each neighborhood card indicates the number of citizen-generated photos and reviews that the user has chosen to save for a particular area of the city. Neighborhood cards can be viewed and shared between multiple users. Cards can be revisited at any time for the addition of photos and reviews.
|
The system allows the user to caption and save these images to share with his/her moving partner at a later time. Based on a user’s caption keywords to these on-site images, the system is able to display local, relevant social media posts that may be helpful to save as well.
|
Back at home, the remote user receives and reviews the on-site documentation and annotations of his partner’s experience.
|
Conclusion
Upon completion of these prototypes I was able to reflect upon their relationship to each other and to the overall purpose for this thesis. First, there is great value in designing for digital maps which can appropriately respond to multiple phases of a decision-making process instead of any single one. To do this, I found that the system must also provide interactive areas that are distinctly not cartographic in nature. In Prototype 1, the user’s preference definition and weighting “space” is an example. This is an area in which the user determines what he/she would like to see displayed in the map “space.” As these two spaces serve related, though distinct purposes, their design must be carefully considered so the user is certain about the function of each. Previous studies during this thesis investigation demonstrated that too much interaction or visualization within and on top of the map space quickly became overwhelming and confusing as to how the user’s preferences controlled the type of data visualization at the city-wide view.
Additionally, there were a number of unanticipated insights that surfaced throughout this investigation. What became a central theme as this investigation progressed was the opportunity for the user to simultaneously understand multiple scales; both in terms of cartographic space and in terms of one’s understanding of preference. Both of these concepts included a generic description: for cartographic space, this description was the general, city-wide pattern visualizations (i.e. concentration and size of graphic markers for each preference); for the preference-definition phase, the system displayed general categorical search terms (i.e. “Food Options”). These hierarchies also included more specific instances. The generalized patterns on the map at the city-wide scale resolved into particular locales at the neighborhood-specific scale, each with unique data characteristics and accompanied by user-generated content. Similarly, each preference category contained a number of specific search terms, which the user is able to redefine at any stage in the decision-making process.
Additionally, there were a number of unanticipated insights that surfaced throughout this investigation. What became a central theme as this investigation progressed was the opportunity for the user to simultaneously understand multiple scales; both in terms of cartographic space and in terms of one’s understanding of preference. Both of these concepts included a generic description: for cartographic space, this description was the general, city-wide pattern visualizations (i.e. concentration and size of graphic markers for each preference); for the preference-definition phase, the system displayed general categorical search terms (i.e. “Food Options”). These hierarchies also included more specific instances. The generalized patterns on the map at the city-wide scale resolved into particular locales at the neighborhood-specific scale, each with unique data characteristics and accompanied by user-generated content. Similarly, each preference category contained a number of specific search terms, which the user is able to redefine at any stage in the decision-making process.