-
Notifications
You must be signed in to change notification settings - Fork 27
Options for symbol
Use case (taken from usecase page)
Those who have a severe speech and physical impairment the use of symbols to represent words is their primary means of communication for both consuming and producing information. Some users communicate through the use of symbols, rather than written text, as part of an Augmentative and Alternative Communication (AAC) system. Symbol users face a wide variety of barriers to accessing web content, but one of the main challenges is a lack of standard interoperability or a mechanism for translating how a concept is represented in one symbol set to how it may be represented in another symbol set.
Examples include:
-
an assisted living home make adult education course and life-skills content. For example, they have content on how to make dinner using a microwave, but they have people who can read different symbols. They need content to be able to work for all their users. sometimes the symbols or pictures used are unique to the user, such as the picture of the actual person phone or cup etc.
-
People who know different symbol sets wish to talk to each other
-
A government agency are making information sheets about human rights and patient rights. they add symbols for lots of different users but they wish people who can read different symbols to be able to read it as well.
-
A large banking site wants people to be as autonomous as possible and use their services. they have augmented symbol references onto their core services.
It should be noted that the users who depend on symbols the most may struggle the most with miss translations, as they have severe language disabilities inferring what was meant by use of an incorrect symbol will not be achievable for many users. This rules out relying on machine learning until it stops making errors.
The task force understands that making a scalable solution may, therefore, make a solution not useable for the earlier use cases. We may need a multi-phase solution so that the earlier usecase are supported sooner.
we probably need
- a standard reference mechanism for each concept common in symbols (such as the BIC number / a url an/or a term)
- a way to associate new symbol(s) with the reference mechanism in the document or content
- a way to associate new symbol(s) with the reference mechanism in a mapping file (phase 2 for scalability for authors if the content provider is using their own symbols?)
- a way to reference a mapping file to use in the content (phase 2 for scalability for authors if the content provider is using their own symbols?)
AT this point we seem to agree
- a standard reference mechanism could be the number user by bliss or the common term for the concept for very comen concepts. This is like using a number for the color in CSS but you can also say "green" or another term for a limited set of very common colours
Map symbol libraries to common concepts, so that symbols can be accurately converted to text in any language and then presented to other users in their preferred symbol library. The ability to personalize a webpage and present symbols instead can help those easily understand the content being provided.
Example: In this example the author would add symbol=”name space references for the image” < span symbol="http://blisssymbolics.org/refnumber/17978" >Example
When there is more than one concept, multiple concepts can be referenced by separating them with white space. The order of multiple concepts should be subject predicate object. Both omissions and multiple triples are allowed.
Note that a unicef consortium is mapping the different open symbol libraries to a reference symbol (at blisssymbolics) to enable this. See the implementations page
code: <span data-symbol="http://blisssymbolics.org/refnumber/17978" >Example </span>
Here we use bliss refrnce nodes. (
Advantages
- Each symbol can be mapped to words or terms. Has been mapped to other concepts libraries
- Bliss is a very big and international - almost 6000 concepts that are in use as unique symbols concepts
- Bliss have a process to add new symbols about every 3 months. Anyone can propose a new concept and hope to have it give a node URI in 3 months
- They address all parts of speech
- They are excited to help
- makes it easy to automatically add actual symbols as well
- licencing creative commons option http://www.blissymbolics.org/index.php/licensing
- BCI currently maintains 9 diverse languages but is in use in many more locations, so it is not very culturally biased.
Disadvantages
- we do not "own" Bliss
- Long URLS
links
http://www.blissymbolics.org/index.php/resources
Use only the refrence number from as used by bliss
code: <span data-symbol="17978" >Example </span>
Here we use bliss refrnce nodes.
Advantages
- We can use any BCI reference number so we have most of the advatages of bliss
- we are not dependent on BLISS. We can keep a document of what each number represents and add to it if we feel a need
- shorter URLS
- if bliss add new symbols our symbols mapping should just work, so long as diffrent symbol sets maintain their mappong
Disadvantages
- slightly harder to automatically add actual symbols as well
code: <span data-symbol="example" >Eg: </span>
Here we use a term for each symbol. we can start off
Advantages
- Terms are easier to read and understand
- We could merge with purpose?
Disadvantages
- harder to automatically add actual symbols as we need to create mapping
- our library of terms will age (bliss keep it maintained). This is a big deal as language on the internet changes fast
code: <span data-symbol="example" >Example </span>
code: <span data-symbol="17980" >Unversity</span>
Here we use a term for each symbol but you can also use the bic reference, so it will grow with bliss,
Advantages
- Terms are easier to read and understand,
- it grows with bliss if we stop maintaining it,
- we can also add more terms if we find bliss stops working well for us, or are missing terms
<span data-symbol="wd:example" >Example </span>
This has a namespace for "WD" such as wordnet.
Note that the sytax may be wrong - we need to work out the right way to word this
Advantages
- Often more terms then bliss
Disadvantages
- So many terms that brake the mapping. For example:
code: <Lemma writtenForm="paternal grandfather" partOfSpeech="n"/>
- will this get mapped to grandfather - harder to automatically add actual symbols as we need to create a mapping
- we may need to maintain the "allowed terms"
- often culturally biased or language biased (wordnet is English)
- some parts of search are NOT supported so we need to augment
- terms are not mapped to symbols
then people can have there own symbols and map to it generically. If we have code inline the mapping is in place.
We can have this work for all tokens
<img href="mysymbolset.com/girl.html/>
we need:
- In the header add a link to the mapping files. Example syntax:
<link rel="personalization" scr="mymappingsfile.html>
- A protocol for overrides so that you can use an image to mean something else - For example - one of the inline options above would still need to be implemented for cases where a word is used differently. Alternitvly we could use CSS cascading rules and allow people to add a higher priority file with overrides and more limited scope using xpointer. Or both.... (see https://lists.w3.org/Archives/Public/w3c-wai-gl/2003AprJun/att-0002/Interoperable_Language_Standard_-_resource_document.htm for ideas)
- A format for storing the mapping in the mapping files. NOTE the syntax is not being proposed now it is just for demonstration purposes. Let us decide if we like the direction and then agree on the syntax. Example syntax (we know it is incorrect)
<symbol for="mysite.org/exampletic.jpeg" value="17980"/>
<destination for="mysite.org/contactus.html" value="contactUs"/>
or some owl/rdf equivalent or a json version or both.... json examplemysite.org/contactus.html{ context:contactUs;
easyLang:"email us or phone us";
symbol:"539" }
Advantages
- essential if the author is actually adding symbols to the content - see Johns email at https://lists.w3.org/Archives/Public/public-personalization-tf/2019May/0002.html
- makes it scalable for other cases - see my email at https://lists.w3.org/Archives/Public/public-personalization-tf/2018Dec/0005.html
- we can extend this proposal to enable all cloud-based accessibility such as alt text, long discriptions and aria.
Disadvantages
- maybe we should focus on this after we have the inline sorted out
- we still need a way to put it inline as sometimes words and symbols are used to mean different things
- people will link to mappings file because it is easy and not address the ambiguity. this will cause people to not be able to use the content. Note that people with cognitive disabilities will often not be able to "hack" the translation and automation mistakes. This is for people with language disabilities so having one in ten words incorrect will make it useless for a lot of the intended audience.
- data-symbol
Symbols are a special type of simplification, where images take the place of text strings - for example a form label of "Address" could be rendered instead as a picture of a house (). Symbols alone however, while graphic elements, should not be called using the <img>
element; rather one of two things should happen:
-
Individual symbol-sets should map their values to known token values, with the expectation that users who require symbols will already have a set of symbols installed on their computer (for use in both web pages, web applications and stand-alone computer programs). There are multiple symbol-sets and icon-sets available, and ideally the end-user will get the symbol they are accustomed to, rather than one dictated by the page author.
For example, here are multiple instances of "home" (house): | | | | - it is not unreasonable to suggest that an end-user, anticipating this may not recognize that it is the equivalent to (or vice-versa). -
However, expecting all users to have a specific symbol set installed may be overly optimistic, and so, as a fall-back authors could suggest a symbol taken from a common symbol set. If this were the case, user-agents that already have a symbol-set installed MUST ignore the author suggestion and render the appropriate pre-mapped symbol. Only when a pre-mapped symbol does not exist should the user-agent then apply the author suggested symbol on-demand:
<label for="bar" data-symbol="URL for the symbol">Your Principle Residence</label> <input type="textarea" id="bar" name="address" data-purpose="street-address">
Example:<label for="bar" data-symbol="http://blisssymbolics.org/refnumber/001">Your Principle Residence</label> <input type="textarea" id="bar" name="address" data-purpose="street-address">
In this example, IF the data-purpose value has been mapped to a specific, installed symbol, THEN the user-agent will replace the<label>
's text value with the pre-mapped symbol. If however the user-agent has been flagged to always render symbols, but an existing symbol does not exist on the user's device, THEN the user-agent will source the referenced symbol provided by the author.
@data-symbol MUST only take a URL reference to the actual symbol, and is used in conjunction with either purpose (form inputs), action (this button does what?), or destination (this link takes me where?)
- Ideally each of the token values used for data-purpose, data-action, and data-destination would also map to a unique symbol (at which point there would be no need to explicitly call for a symbol, as the user-agent would already have symbols related to all of those important functions, and would be configured to do the transformation on all content).
(Question to be determined: should the specification explicitly re-state all of the token values used elsewhere here?) - Page authors can none-the-less suggest a specific symbol with the understanding that it is only a hint, and that most users who require symbols for communication will already have one or more sets installed on their machine. In this case, the hint is furnished via a URL to the specific symbol being used