SOMU - A semantic model API

Lawrence Meckan

Absalom Media

The state of the web

• Websites do not have any semantic understanding between each other. They remain islands of semantic behaviour.
• Minimal code sharing regarding semantics
• Screen scraping doesn't grab the semantic elements - it only focusses on data
• Design piracy reflects the state of the web - not only are people lifting copyrighted template material, but also attempting to replicate semantic structures (badly)
• CMS designs tend to vendor lock in a specific semantic model or way of deploying semantics
• No CMS allows accessibility solutions to affect both core and third party work as one ubiqutious way of looking at output
• The problem of accessibility exists across 95% of the CMS marketplace.
• CMS products are a smaller market than the web industry itself so the actual figure may be even higher, based on John Allsopp's work.
• The web semantic model doesn't need to be defined by macroformatted JS engines e.g. Macroformat JS
• Open source content management sucks.
  

  Think how much easier your CMS will be if I could simply say, "I want these features presented in this order," and then apply a stylesheet that does all of the presentation.

• CMS products have not iterated the semantic model in 3 years - Mambo and Joomla still rely on the same principles to deal with semantics as 3 years ago.

SOMU - Aims

SOMU - What is it?

• A library documenting semantic behaviour
• Some form of XML/XLST leveraged by device-agnostic code to replace overall semantic structures
• Not semantic areas themselves, though data using the semantic model must always be localised.
• Not microformats, though microformats should be able to be used within it
• A new solution to the old issue of web accessibility

SOMU - Basic principles

Web page structure can be standardised

• Every page has some form of heading
• Every part of the 'content' output conforms to some layout of some kind
• Every component, module and plugin (CMP) for Joomla and Mambo follows a similar path, yet can't leverage any accessibility benefits
• Similar design principles have appeared in other CMS systems - the same functionality turns up in Drupal or Wordpress
• Table structures need to follow certain principles to be web accessible, something most WYSIWYG editors fail at
• Templates always take care of the overall structures outside the core and the editors
• So why not inside the core of a CMS?

SOMU - Brainstorming

To deliver the solution needed:

• Use some form of simple templating engine
• Smarty and PatTemplate are too heavy for this job. Need something lightweight ? What ?
• XML/XLST templates have been used with success before in developing for CMS products
• Pseudo-code ideas for input/XML/XLST: <?xml version="1.0"?> <page> <heading> h3 class="entry-title" </heading> <article> td class="contentpaneopen" </article> </page>
• Pseudo-code ideas for output: <article> <heading> Blah </heading> </article>

SOMU - More brainstorming

• Stack each semantic area inside another, to allow precedence and inheritance
• Iterate it down to the smallest semantic element on a page
• Areas could include: blog, 2 column, 3 column, article, leading page, 4 column
• Sub-blocks could include: heading, footer, summary, publish date
• XML/XLST based templating has been around for a while. Why not turn it into an open API?
• Able to level the playing field across the web by providing it as an open XML API

SOMU - Directions

• Simplify and standardise
• Drive the model to the user - give them the choice of what semantic model they need
• Remove or mitigate device-dependent structures across multiple CMS players
• Open standardisation of the semantic web
• Potential ability for 'hot-plugging' outside the CMS marketplace as well
• Each CMS has a slightly different semantic model. Share them in line with Web 2.0 'open standards'
• Every other part of content can be manipulated though APIs. Semantics is uncharted territory.
• 'Hot-plug' into CMPs for both core and 3PD structures and all get the same benefits.
• Mass customise the semantic model to the point of spawning a cottage industry based on a sharable XML file

SOMU - Benefits to the CMS marketplace

• Semantic model allows people to switch between CMS products with greater ease and less visual or accessible disruption
• Open API - open source, open standards benefits
• API can grow and mature over time allowing new layouts and semantic structures to be shared

SOMU - Benefits to developers

• Semantic model can be pre-build / 'hot-plugged' to suit
• API is XML based - use it anywhere. Why not syndicate it?
• Potentially less code, with better structure and achievable output
• Potential for developers to market their code as accessible out of the box
• Conforms to "Don't make me think" principle of design
• Someone else ends up resolving your accessibility issues
• Developers gain the benefits of accessibility experts working on the solutions for them
• Write-once, read many - based solution.
• As Jeremy Keith talks about, API's are sexy

SOMU - Benefits to designers

• Semantic model can take care of accessibility beyond mere mechanical tools/validation
• Don't have to think - just 'hot-plug'
• Designers can then focus on visual messages instead of worrying about code
• It's like the CSS Zen Garden, but using a semantic model, instead of CSS, to convey variations in design.

SOMU - Benefits to end users

• Sites customised quicker - lower cost for TCO
• They don't have to worry about accessibility - semantic model manages that
• They can potentially share the same semantic model across multiple domains and sites, using different products (CMS / AMS / CRM)

SOMU - Benefits to web industry

• Simple solution to a complex problem
• Sharable due to open standards (XML/XLST)
• You start mashing up the entire design process
• Your semantic model can be decided before you even touch a line of code or involve a designer

SOMU - Benefits to accessibility providers

• Can determine the marketplace for semantics
• Greater voice for accessibility issues through sharable open standards
• Semantic model 'builders' may continue to evangelise the issues through delivery of a solution
• Ease of use to implement relevant standards - ATAG, WCAG2, 508.. Pick and choose the template you need
• All involved can compete and compare semantic models - evolving the rest of the industry into a truly semantic web

SOMU - Gameplan

• Stage 1: Document the semantic model matrix - Drupal, WP, Joomla 1.0.11 & 1.5, plus any current accessible "solutions" Completed
• Stage 2: Standardise an XML file that can interface to all of them Currently in progress
• Stage 3: Implement core and third party plugin examples for each CMS.
• Stage 4: Deploy and educate to the wider web industry