atbcb / ICTTestingBaseline

Baseline tests reduce ambiguity, increase consistency of results, and emphasize testing of the methods and techniques that can reliably meet the Section 508 ICT Accessibility requirements, given the current state and compatibility of underlying technologies.
https://ictbaseline.access-board.gov
Other
45 stars 23 forks source link

Adding Testing Criteria Specific to Digital Geographic Maps #477

Open frastlin opened 3 months ago

frastlin commented 3 months ago

Overview

Currently, digital geographic maps are evaluated using section 6. Images, but digital geographic maps are too complicated to be combined under images without explicit evaluation criteria. Digital geographic maps are complex multicomponent elements that require an entire evaluation of their own, above the general website evaluation. In particular, WCAG 1.1:Non-text content requires a more nuanced definition. Having requirements specific to digital geographic maps will reduce ambiguity of what to evaluate on geographic maps, and ensure evaluation criteria are clear and consistent across results so that the section 508 standards are applied correctly to this complex component.

An Accessibility Conformance Report was done on the top 11 digital map tools, and the below criteria are the most common failures.

Section 1.1.1:Non-Text Content Applied to Maps

Section 1.1.1:Non-Text Content states: “All non-text content that is presented to the user has a text alternative that serves the equivalent purpose”. The problem is that “equivalent purpose” is somewhat ambiguous and is often misinterpreted. In academic literature the framework of landmark, route, and survey knowledge acquisition is used, and that is what I propose be the refined framework for section 1.1.1:Non-Text Content applied to digital geographic maps.

Traditionally, the practice of making maps accessible is to step back and look at the “point” of the map. Is the alternative that’s presented conveying the primary objective of the map? This practice can be observed by blog posts written by prominent digital accessibility experts and documentation by government agencies: https://www.tpgi.com/accessible-digital-map-experiences/ https://mn.gov/mnit/about-mnit/accessibility/maps/static.jsp https://github.com/DEFRA/design-standards/blob/master/Maps.md

This practice is not compliant with section 1.1.1, as following this method only has the “main purpose” conveyed when the criteria states “equivalent purpose”. If the “main purpose” of a university campus map is to show roads on campus, it does not help the map user needing to manage garbage cans on campus. A visual campus map would have garbage cans, but since they’re not the “main purpose”, the map user accessing the text alternative would not be able to use the map for the “equivalent purpose” that the original map could have been used for. Using the survey knowledge framework would facilitate this “equivalent purpose” though.

Siegel et al. published a seminal work in 1975, titled “The development of spatial representations of large-scale environments” defining the framework of landmark, route, and survey knowledge for evaluating spatial knowledge acquisition from geographic maps. This framework has subsequently been used in numerous studies across maps in visual, auditory, and tactile modalities including: https://cartographicperspectives.org/index.php/journal/article/view/cp75-ishikawa-takahashi/html https://www.tandfonline.com/doi/full/10.1080/13875868.2021.1969401?casa_tok https://www.sciencedirect.com/science/article/pii/S0272494407000734 https://icad2022.icad.org/wp-content/uploads/2022/06/ICAD2022_27.pdf https://www.frontiersin.org/articles/10.3389/fnhum.2014.00223/full https://www.sciencedirect.com/science/article/pii/S0272494406000594

This framework is rigorous, quantifiable, and easy to use.

Definitions and criteria to evaluate each element of landmark, route, and survey knowledge:
  1. Landmark, which is the sensory characteristics, name, type, shape, orientation, and size of a particular feature or location on the map;
  2. Route, how two landmarks connect including distance, direction, legs of the route, and shape of the route;
  3. Survey, which is the overall understanding of how all the points, polygons, and lines relate through their distance, direction, shape, size, orientation, and most importantly, general layout on the map.

Using this framework of spatial knowledge allows a quantitative evaluation of the spatial information in a map. Additionally, there needs to be full access to the numeric or other information that may be overlayed on the geographic features. If a text alternative is missing any of these elements, it does not serve the “equivalent purpose” of a map, and is not section 508 compliant.

Section 1.4.11:Non-text Contrast Applied to Maps

1.4.11:Non-text Contrast states: “The visual presentation of the following have a contrast ratio of at least 3:1 against adjacent color(s): User Interface Components; Visual information required to identify user interface components and states, except for inactive components or where the appearance of the component is determined by the user agent and not modified by the author; Graphical Objects; Parts of graphics required to understand the content, except when a particular presentation of graphics is essential to the information being conveyed.” Unfortunately, for thematic maps showing more than 3 categories of colors, it’s physically impossible to comply with this criteria as written. Recommended compliance criteria should be as high of contrast as possible color scheme, coupled with patterns to distinguish between the values. The pattern criteria works for colorblind users, and the high contrast colors work for low-vision users.

Section 2.1.1 Keyboard Applied to a Map

Section 2.1.1 Keyboard states: “All functionality of the content is operable through a keyboard interface without requiring specific timings for individual keystrokes, except where the underlying function requires input that depends on the path of the user's movement and not just the endpoints.” This includes keyboard accessibility for all features (e.g., points, polygons, and lines) on the map. For map creation tools, drawing with the mouse or touchscreen may be dependent on a path of the user’s movement, but it’s possible to place points and vertices with the keyboard.

Proposed Content for Section 508 ICT Testing Baseline: 25. Geographic Maps

Accessibility Requirements
Test Method Rationale
Limitations, Assumptions, Exceptions
25.A Test Procedure for Maps

Baseline Test ID: 25.A-Maps

Identify Content

Identify a collection of points, polygons, and lines and possible associated controls that comprise a map, whether it’s an image, interactive, in a document, and or in an iFrame.

Test Instructions
  1. Check that the following are true [SC 1.1.1]: a. It’s possible to identify the sensory characteristics, name, type, shape, orientation, and size of all features (e.g., points, polygons, and lines) on the map through text. b. It’s possible to understand the distance, direction, legs of the route, and shape of the route between any two or more features (e.g., points, polygons, and lines) on the map. c. It’s possible to obtain the distance, direction, shape, size, orientation, and general layout of and between all features (e.g., points, polygons, and lines) on the map. d. All numeric values (if the map contains numeric values) are accessible and connected to all the features (e.g., points, polygons, and lines) associated with the numeric values. e. Logos have alt-text in the “alt” attribute. f. Icons (e.g., “open map in new window”, zoom in, zoom out, close, copyright, etc.) have text labels or an “alt” attribute with text describing the functionality of the control. g. Decorative images or text are hidden from screen readers with aria-hidden.
  2. Verify that information (such as a control's state), structure, and relationships conveyed through presentation can be programmatically determined or are available in text [SC 1.3.1]. Look for the following: a. There’s a “disabled” attribute on the buttons that can be disabled (e.g., zoom, or reset orientation). b. There’s an aria-pressed attribute on the buttons that can be activated or deactivated (e.g., locate me, and full screen). c. The entire map is wrapped in a labeled region landmark so screen readers and other assistive technology know what is in that area. d. The Scale bars/rulers are understandable non-visually, or they are hidden from the screen reader. e. For thematic maps, all numeric values and ranges are accessible to a screen reader and comprehensibly associated with their related features (e.g., points, polygons, and lines). f. All feature properties (e.g., name, type, information card) are possible to activate in the text and other modal versions of the map. g. Spatial relationships (e.g., route and survey knowledge) can be ascertained through text.
  3. Verify that the visual presentation of text and images of text has a contrast ratio of at least 4.5:1 [SC 1.4.3], except for the following: a. Large Text: Large-scale text and images of large-scale text have a contrast ratio of at least 3:1; b. Incidental: Text or images of text that are part of an inactive user interface component, that are pure decoration, that are not visible to anyone, or that are part of a picture that contains significant other visual content, have no contrast requirement
  4. If there are more than three discrete ranges being represented and it’s impossible to meet [SC 1.4.11], patterns are provided as an optional alternative.
  5. Ensure all map functionality can be accessed using only a keyboard [2.1.1]. Look for the following: a. The map display is pannable using the keyboard. b. Map controls and information are possible to access without using hover with the mouse. c. Drag and drop controls can be operated via the keyboard. d. All features (e.g., points, polygons, and lines) can be focused using a keyboard (preferably not just through tab.) e. If there is map creation, points, polygons, and lines can be created anywhere on the map using the keyboard.
  6. Check all the following are true [SC 2.4.7]: a. It’s possible to sequentially navigate all components using the tab key (and operate all controls). b. Check that scripts are not used to remove focus and that each component is highlighted with a visible focus outline.
  7. The lang attribute is set on all webpage and iFrame components, and the specified language code matches the language of the content (which may change due to language negotiation, based on system settings) [SC 3.1.2].
  8. For all user interface components, look for the following [SC 4.1.2]: a. The name and role can be programmatically determined, including for the map itself, any parent iFrame, controls all have an accessible name, and controls have the correct role; b. States, properties, and values that can be set by the user can be programmatically set; b. and notification of changes to these items is available to user agents, including assistive technologies. c. If the map is embedded in a document that does not have an alternative version, there is a link to a Conforming Alternate Version.
Test Results

If any of the above checks fail, then Baseline Test 25.A- Maps fails.

Advisory: Tips for streamlined test processes
WCAG 2.2 Techniques

The following sufficient techniques and/or common failures were considered when developing this test procedure for this baseline requirement:

frastlin commented 2 months ago

Here are passing non-visual map examples, let me know if you'd like some examples of failing maps as well:

Detailed Text Description

Although this level of text description is rare, let’s try and hit all the elements of spatial knowledge for a few states:

Evaluation of Detailed Text Description

Interactive Alt-Text

Here is an Audiom map of COVID Cases over the U.S. showing interactive alternate text. Here is a video showing the Audiom interactive alternate text usage.

Evaluation of Interactive Alt-Text

Here is a video requiring headphones showing a campus map with interactive alternate text as well.

KMSOC commented 2 weeks ago

If we are solely limiting discussion to what we should required in alt text for maps, below are several thoughts:

  1. Alt text should follow the guiding principle for any image: If I removed the map and just had the information in the document and the alt text, would all the critical information being communicated by that map still be fully available? What critical information would I lose if I couldn’t perceive the map? If there is information loss without the image, there may be more work to do to ensure that the map is accessible to everyone.

    • Our challenge is still related to complex maps that are meant to be exploratory or navigational. In these cases, I would suggest attempting to understand the author's intention. If the author did not intend the map to be used in that way, it should be acceptable to have alt text that matches the author's intention (also understand that the tester may not be able to know the author's intention)
  2. Where possible, an additional text alternative for all data and information is provided

  3. Where a text alternative to full describe map content and context is not possible, include, if possible, an alternative means to access content such as a POC, link to google map, or other interactive means that will provide an accessible alternative

    • Understood that a POC may not always be an option nor reasonable option
    • Understood that a google map or other alternative map will not be available or possible
    • There are still gaps here as there may be many scenarios where it isn't possible to achieve any of the above. Does that mean that complex navigational maps will always fail until an accessible alternative is provided?

We also had a suggestion to split out minimum requirements based on the type of map. Some types of maps (in more plain language) are:

  1. Data visualization maps (e.g., topographical, country, economic, weather, other data or theme-based representations)
    • These most likely have data driving the visual representation and often can also express the data in accessible tables
  2. Navigational maps (road maps, campus maps, maps used for directions and navigation)
    • These map be simple or complex. Simple maps may be easily summarized in alt text. Complex maps include myriad starting and stopping points which are impossible to describe in entirety.
  3. Property maps (often hand drawn due to land surveys)
    • Specific to show land delineation and may be easily summarized in text format.
  4. Historical maps
    • Broad grouping of older maps where author's purpose may not even be able to be discerned. Text may also be faded/missing.
frastlin commented 2 weeks ago

Hello, The above response seems to reiterate the current best practice presented by places like the state of Minnesota. Current best practice is not systematic, quantitative, rigorous, reproducible, or evidence based. The current best practice is harmful to people depending on the alternate text for viewing maps because it is not an “equivalent” access to the content. Blind people are calling for these practices to be updated to something that can allow for professional blind cartographers to exist which the current best practice does not allow. The proposed test is quantitative, rigorous, reproducible, systematic, and based on decades of research in the cartographic community.

  1. The guidance for 1.1.1 states “All non-text content that is presented to the user has a text alternative that serves the equivalent purpose”. WCAG does not mention “critical information”, it only mentions “equivalent purpose”. If the author is only providing “critical information”, that is an author interpretation that could lead to any kind of alternate text of any granularity. It’s very important that equality is not completely subjective. This is why this issue was created, because current alt-text for maps follows the practice of identifying the “purpose” or the “critical information” as identified by the author. This could lead to alt-text for the eclipse map of the lines of totality saying something like: “This map is of the lines of totality across the contiguous United States. The line of full totality begins at the southern border of Texis, goes northeast past both San Antonio and Austin, continues northeast to Dalis, then enters Oklahoma. The line of full totality covers the eastern edge of Oklahoma and the western edge of Arkansas, just missing Littlerock. The line then travels northeast, covering the central part of Missouri, then slight northeast into Illinois. In Illinois, the line of full totality goes east into Indiana and covers all but the southern edge of Indiana. The line of full totality then travels east into both Michigan and Ohio. The line then travels northeast into the northwestern edge of Pennsylvania, and into New York. The line of full totality passes over Buffalo, and east into Vermont. The line then goes northeast over the northwestern corner of New Hampshire and into Maine. The line then travels through the center of Maine, covering all but the very northern and very southern edges. The line of full totality then continues northeast.” This alt-text follows current best-practice, but it’s missing the majority of the information from the above map. It is not systematic in presenting including distance, shape, size, and general layout of the points and polygons. This alt-text is missing everything about the shapes, sizes, orientations, and distances of the state polygons, the exact locations of the city points and how far the city points are from the other states or city points, and there’s no information about the lines of 95% totality, 75% totality, or 50% totality. This alternate text fails to give the “equivalent purpose” to the visual map because it is missing all the above information. Using the interactive alt-text eclipse map I was able to write this text description, and I could have added all the missing detail. I’m completely blind. The above interactive alternate text map provides distance, direction, shape, size, orientation, and general layout of points, polygons, and lines. The above interactive alt-text map shows all lines of totality, as well as states without any lines of totality. The remainder of the comments assume it’s possible to fully convey geometries through alt-text. a. The “author’s intention” is different than the “equivalent purpose”. 1.1.1 doesn't mention an author’s “intention”. “intention” is subjective and is not “equivalent”. If the author did not intend to show geometries, they would have left them off the map, which means there is no longer a map and this test is not applicable.
  2. “Where possible” is also not mentioned as an exception to 1.1.1. Section 1.1.1 states for “charts, diagrams, audio recordings, pictures, and animations, text alternatives can make the same information available in a form that can be rendered through any modality”. It is always possible to provide a text alternative to a map that follows these guidelines. If you can give me some examples of when alternate text following these guidelines is not possible, we can discuss those. a. This COVID statistic map from the CDC does not convey any distance, direction, shape, size, orientation, or general layout of points, polygons, or lines. True, the data table is overlaid on the geographic map, but the visualization has the geometries of the counties. This thematic map from the CDC overlays the data table content on the geometries visually, so the text alternative should have the geometries as well. Just looking at this table, can you answer “What county bordering “CA, Santa Clara” has the highest number of vaccinated people? No. Can you answer this question when looking at the visual map? Yes. This is a clear violation of 1.1.1. I don’t have an example of an interactive alt-text map showing counties, but here’s an interactive alt-text showing COVID statistics over the United states. Can you answer “What state bordering California has the highest number of cases per one million through text?” Yes, Arizona.
  3. “possibility” is different than "equivalent purpose" mentioned in 1.1.1. It is possible to have a fully accessible alternative for any map.
  4. Can you please give me an example of a map where the geometries are not critical to the map representation? The entire definition of a map is that it’s a symbolic spatial representation. If geometries are removed, what’s left is a table. Removing geometries in the alt-text does not serve the "equivalent purpose" as the visual map that includes geometries, so does not meet 1.1.1. a. Data visualization maps: These maps have tabular data overlayed on them, but the reason why tabular data is overlayed on geometries is to convey spatial relationships between the geometries and spatial data. How is it possible to understand spatial relationships only using a table? How can a scientist understand the spread of ice melt over the Atlantic ocean with only a table? They can’t. b. Navigational maps: As mentioned previously, a “summary” is different than the “equivalent purpose”. When reading section 1.1.1, "summaries" are only mentioned in relation to a data chart. According to G95 “summaries” are only applicable when meeting 1.1.1. if they are “a short text alternative that briefly describes the non-text content. (A long text alternative is then provided using another technique such that the combination serves the same purpose and presents the same information as the original non-text content.)”. There needs to be a clear, systematic way of evaluating the longer alternate text that’s provided, and just saying “give a summary” is inadequate. The purpose of this issue is to clarify that “equivalent purpose” really means the spatial aspect of the map. Please provide an example map that’s “impossible to describe in entirety” and I’ll demonstrate how to do it. It’s possible to create alt-text that describes the entire spatial features of the U.S. down to the subatomic level. The difficulty is creating, indexing and easily understanding this information. The methods of interactive alt-text and long-form alt-text in post two are just two ways to do it. Additionally, a campus map is used for activities other than navigation. Information Technology professionals may use the campus map to identify locations for Wi-Fi hotspots. Universities may indicate to graduating seniors where to line up for commencement. Some campus maps even show real-time occupancy levels of computer rooms in buildings, which starts putting them in the realm of thematic maps. c. Property maps: They should be vectorized into geojson (or something similar) and instead of summarized, described with distance, direction, shape, size, orientation, and general layout of points, polygons, and lines. All the interactive alt-text method requires is a geojson file of the data to render interactive alt-text with any desired granularity. d. Historical maps: They should also be vectorized and turned into geojson (or something similar). Where text is faded or missing, that should be indicated in the property of the geometry. If geometries are partial or unclear, then that should be indicated in the property of the geometry that is understandable.

Maps can be categorized into thematic or referential, but the one thing they all have in common are geometries. There are thousands of other types of maps than those listed above, including utility, service outage, emergency exit, oceanographic, Steller, fictional, multivariate, time-sequenced, etc. and it’s important to have guidance that encompasses all map types. Having separate rules for each type of map would make the rules too lengthy. All maps convey geometries, so I'm arguing that geometries should be the foundation for any accessible map. There needs to be clear guidance for conveying geometries in alternate text, and that’s what this proposed test provides.

There seems to be two underlying misguided assumptions in the current best practice: (1) alt-text is severely limited in what spatial content it can convey; and (2) Spatial content is inherently visual. The interactive alt-text method presented by maps like this Eclipse map shows it’s possible to convey detailed landmark, route, and survey knowledge. This includes distance, direction, shape, size, orientation, and the general layout of all points, polygons, and lines on a map. This data can be conveyed at the same fidelity as the visual map. It only takes a few minutes to generate this level of detail from a geojson file. Not only is it possible to convey this level of detail, but it’s fast and easy. This method of interactive alt-text is just one way to convey this information, I’m sure there are numerous other ways to convey this level of detail as well. There’s nothing visual about geojson. If a fraction of effort was put into text representation of geometries as is put into the visual representation of geometries, I’m sure there would be a number of easy ways to convey this information. As it stands, we have both the interactive alt-text method above, and the long-form alt-text described in post two.

The blind and low vision community is unhappy with the current alternate text practices for representing maps. Only conveying “summaries”, “critical information”, “author’s intention”, and “Where possible”, is keeping blind people from independently participating in professions requiring the use of digital geographic maps. The “minimum requirements” of the alt text need to be at the level so the alternate text user can use the map for the “equivalent purpose” as their sighted counterparts. This does require a paradigm shift away from only considering alternate text created by hand and static alt-text. On July 7, 2024, the National Federation of the Blind passed a resolution calling for maps to convey the level of detail outlined in this issue. This resolution represents the voice of over 50,000 blind people. Over 50,000 blind people are saying the current method of alt-text on maps is inadequate.

The authors of the WCAG were clear that the alternative text representation needs to show an “equivalent purpose” to the visual representation. This means that blind people should be able to answer any question a sighted person can answer about a map. Using this interactive alt-text eclipse map, or systematic long-form alt-text as described in comment two, blind people can.

goodwitch commented 3 days ago

I just wanted to add a comment about this statement (from the original post at the top of this thread):

For thematic maps showing more than 3 categories of colors, it’s physically impossible to comply with [SC 1.4.11] as written. Recommended compliance criteria should be as high of contrast as possible color scheme, coupled with patterns to distinguish between the values.

So, I helped write SC 1.4.11, and there is a creative way to solve this contrast problem. I call it the oreo method, or the outline method. Instead of having map items touch each other directly, you can outline each item with one of the following:

The outline can be white, or black or pick whatever color works best for your color scheme. When things get super complex, you can even go with what I call the "oreo outline"...which is a triple line outline...where the online is a black outline (black cookie) surrounded by a white outline ( white frosting), surrounded by a 2nd black outline (2nd black cookie). With the oreo cookie technique...you can distinguish any item at 3 to 1...because you controlled the background. If black doesn't work as the cookie layer...you could flip it to white cookie, black frosting, white cookie.