[[Accessibilità/Chartability]] is an accessibility evaluation system specific to data visualizations and interfaces which aims to help practitioners answer the question, “how accessible is my data visualization?” Is a consistent and unified methodology for designers to evaluate the accessibility of their work across the broad spectrum of disability considerations. ## Keywords - information-rich systems - Web Content Accessibility Guidelines - Data experience ## Introduction Accessibility is the practice of making information, content, and functionality fully available to and usable by people with disabilities. - practitioners need to be able to identify accessibility barriers. - ==evaluating the inaccessibility of complex data systems can be a daunting and often expensive task== - State-of-the art automated com- pliance checkers only find 57% of accessibility errors, meaning accessible experiences must still be manually designed and checked for quality - research at the intersection of data visualization and accessibility has yet to meaningfully permeate data visualization tools and communities and primarily focuses on blindness and low vision ## Existing Work in Data Visualization and Accessibility ### Research Advancements in Data Visualization and Accessibility - When we asked “What do we mean by data visualization accessibility research?” we found that ==nearly all topics of study were vision-related== - [[Accessible Visualization. Design Space, Opportunities, and Challenges|Kim et al]]. found that 56 papers have been published between 1999 and 2020 that focus on vision-related acces- sibility (not including color vision deficiency), with only 3 be- ing published at a visualization venue (and only recently since 2018)[^1] [KJRK21] - We have found 2 papers that engage cognitive/neurological disability in visualization and 1 student poster from IEEE Vis, which are all recent (specifi- cally intellectual developmental disabilities [WPA∗21] and seizure risk [SB20,SSB21]) - there is no research specific to low vision disabilities (not blindness or color vision deficiency) unless conflated with screen reader usage in data visualization. ==Blind and low vision people are often researched together, but in practice may use different assistive technologies (such as magnifiers and contrast enhancers) and have different interaction practices (such as a combination of sight, magnification, and screen reader use)== [SHZA16] #### Topic of research Since the 1990s, the most prominent and active accessibility topic in visualization has been: - color vision deficiency - tactile sensory substitutions (aged researches) - Sonification - screen reader data interaction techniques - chart descriptions [LS22] chart descriptions are preferably between 2 and 8 sentences long, written in plain language, and with consideration for the order of information and navigation (Jung et al. [JMK∗22]) Other work at the intersection of accessibility and data visualization - automatic or extracted textual descriptions - haptic graphs and tactile interfaces (These research projects produce artifacts that are high-cost for individual use) ### Accessibility Practices in Data Visualization Tools and Libraries - Libraries on data visualization have broad accessibility functionality built in, but their documentation is technically specific to their implementation (Highcharts, Visa Chart Components (VCC), Graphics Accelerator in SAS) - While these relatively accessible libraries and tools can be helpful for inspiration, their specific techniques and guidance materials are not easily transferrable to other environments or applications where data visualizations are created. - More established visualization libraries like matplotlib, ggplot2, d3js, R-Shiny, and Plotly have left most accessibility efforts to de- velopers, with varying levels of documentation and difficulty in- volved. None of these major tools have a broad spectrum of accessibility options built in and documented. Accessibility is still an afterthought in data visualization and ad-hoc, specific solutions proposed have not led to widespread improvements. ### Accessibility in practice ==Accessibility in practice is largely motivated by standards work or assistive technology== - tactile and braille standards are robust [BAN10], but have limited transferability to digital contexts currently In digital contexts, the most influential body for accessibility is the World Wide Web Consortium’s (W3C) Web Accessibility Initiative (WAI). WAI’s ==Web Content Accessibility Guide- lines== (WCAG) influence accessible technology policy and law for more than 55% of the world’s population. ## Making Chartability ==**Chartability approaches accessibility as a scale, not a state.**== How accessible a Data Experience is, is determined by how few failures it contains. Even the absolute best Data Experience may contain several failures, even after remediation. Chartability è un set di 45 euristiche che chi costruisce una data experience deve verificare per misurare il livello di accessibilità del proprio lavoro. Alcune euristiche prendono ispirazione dai principi di accessibilità delle WCAG (POUR) mentre altre sono state suddivise in tre principi aggiuntivi di accessibilità, definiti da Evalsky: Compromising, Assistive e Flexible (CAF). ### Compromising Euristics based on providing alternative, transparent, tolerant, **==information flows==** with consideration for different ways that users of assistive technologies and users with disabilities need to consume information. Compromising challenges designs that only allow access to information through limited or few interfaces or processes. - provide information at a low and high level (such as tables and summaries) - transparency about the state of complex interactions - error tolerance - data structures can be navigated according to their presentation. ### Assistive Assistive focuses on the labor (**==amount of effort==**) involved in access. Assistive heuristics ensure that both visual and non- visual data representations add value for people with disabilities. - data interfaces must be intelligent and multi-sensory in a way that **==reduces the cognitive and functional labor required==** of the user as much as possible. The Assistive principle focuses on what Swan et al. refer to as “adding value” [SPPW] and what Doug Schepers meant by “data visualization is an assistive technology” [Scha]. ### Flexible Designs must not be rigid in their opinions and ability assumptions and should be designed to be moldable by and adaptive to user needs. [WKG∗11, Lad15] Flexible heuristics focus on robust user agency and the ability to adjust the Perceivable and Operable traits of a data experience. - The preferences that a user sets in lower-level systems must be respected in higher level environments. ## Using Chartability ### Visual testing In order to evaluate contrast, often a combination of automatic (code-driven) and manual tooling is performed - Content is only visual - Contrast - Color vision deficiency - Color alone isn’t used to communicate meaning - Text size, readability and spacing - Gli elementi significativi possono essere distinti l'uno dall'altro - Evitare il rischio di crisi epilettiche - No lampi rossi e animazioni non richieste ### Keyboard probing If a data interface contains interactive element, those elements (or their functionality) must be able to be reached and controlled using a keyboard alone. - Instructions or cues should always be provided ### Screen reader inspecting Screen readers, unlike more basic keyboard input devices, read out content that is textual (including non-visual textual information like alternative text). Using a screen reader to audit is generally the hardest skill to learn. - All valuable information and functionality in a data experience should be available to a screen reader - individual variables about a mark - whether that mark is interactive - a mark status updates that reflect context change provide alerts - statistically and visually important areas of the chart are explained ### Checking cognitive barriers Avoid interpretative issues - Charts must have a visually available textual explanation provided that summarizes the outcome - use of plain language - explanation of specific terms - clarity of all available text - charts have basic text that provides a visually-available textual description and takeaway ### Evaluating context - information flow - Possibily to toggling high contrast modes - Visual in not too dense or complex - there are both high and low level representations of information available ## Further considerations Chartability is going for above compliance and focusing on a good experience. > Access is an Experience, not just Compliance Automations and tooling would help novice practitioners perform this work faster and with more confidence Elavsky made Chartability openly available on Github [^2]. As new research and practices emerge and more community members get involved, Chartability will become an evolving artifact of consensus similar to existing standards bodies. “We want Chartability to become a living, community-driven effort that will adapt and grow as more resources, tools, and research become available.” ## Risorse Per una lista dettagliata e ordinata delle 50 euristiche: [Chartability Notion database](https://lush-vault-6ea.notion.site/Chartability-239695360ccd4982a64339694f6da647?pvs=4) - [How accessible is my visualization? Evaluating visualization accessibility with Chartability. (Official paper)](https://www.frank.computer/chartability/) - [Chartability website](https://chartability.fizz.studio/) - [Intro to Chartability (data.europe.eu)](https://data.europa.eu/apps/data-visualisation-guide/intro-to-chartability) [^1]: “[[Accessible Visualization. Design Space, Opportunities, and Challenges]]”, Kim et al, 2021 [^2]: [Chartability](https://chartability.fizz.studio/)