This page summarises RQTF's activity around research into knowledge domain-specific content and notation.
- 1 Background
- 2 Literature Reviews
- 3 Emerging findings from literature reviews
The Accessible Platform Architectures (APA) Working Group is undertaking an effort to develop strategies for enhancing support in W3C technologies for the accessibility of symbologies and graphical conventions (e.g., diagrams) that occur in specific knowledge domains. These domains include mathematics, chemistry, physics, music and linguistics; however, it is anticipated that other disciplines will also become relevant to this work.
For a statement of the problem, and for further discussion, Knowledge domain accessibility (issue #9) should be consulted.
Emerging findings from literature reviews
Findings from the literature reviews which are relevant to the further development of W3C technologies are summarized below. Several themes have been discerned from the research reviewed by the Task Force.
General Accessibility Requirements
- Spoken presentation (supporting a plurality of alternative speaking styles).
- Automatic summarizing of notation (e.g., use of variable substitution to convey the structure of a mathematical expression, where the entire expression is summarized first, followed by a reading of the individual subexpressions).
- Synchronized highlighting of content (including notation) as it is spoken.
- Visual reflow of content (including, e.g., mathematical notation) upon enlargment.
- In support of reflow, line breaking that respects the semantics of the notation.
- Braille output (in a variety of braille codes, as appropriate to the notation and the codes needed by the user).
- Linearization of two-dimensional structures (e.g., tables, matrices), in addition to providing a navigable, two-dimensional representation.
Preservation of Semantic Distinctions Required to Support Mutliple Renderings
It is clear from the literature that markup languages used to represent domain-specific content (e.g., presentation MathML and SVG) do not preserve all of the semantic distinctions required to automate the rendering of such content in alternative forms. For example, the spoken presentation of mathematics benefits considerably from distinguishing between function application and multiplication of a parenthesized expression in algebraic notation; however, this distinction is not captured in presentation MathML. Further, presentation MathML (in contradistinction to content MathML) is the format generally created by authoring tools - including document conversion systems.
The strategy typically adopted by researchers is to build heuristics that attempt to infer the missing semantics from the actual markup. The results of this analysis are not necessarily accurate. It has been suggested in discussions of this topic that metadata which more precisely specify the subdomain (e.g., mathematical subdiscipline) of the material would improve the reliability of these heuristics. For example, knowing whether the content was concerned with elementary algebra, linear algebra or set theory would facilitate heuristic interpretation of vertical bar symbols occurring in mathematical notation (where the vertical bars enclosing an expression could respectively represent, for example, the absolute value of a real number, the norm of a vector, or the cardinality of a set, according to context).
In the case of presentation MathML, it is also possible to construct alternative, but semantically equivalent representations of the same expression. For this reason, tools described in the literature for rendering presentation MathML as speech, or in various mathematical braille codes, generally convert the MathML to a canonical form prior to carrying out further processing.
The following navigational needs are evident in the literature, where the discussion focuses largely on mathematical content.
- Support for navigating by subexpression.
- Support for moving to the beginning/end of a matching pair of delimiters (parentheses, for example).
- The option to collapse/expand subexpressions.
- Cell-by-cell navigation (up, down, left and right) within tabular structures - for example, tables, matrices, graphs represented as adjacency matrices.
- Navigation between tabular structures - the actual requirement is probably to retain one's place in a tabular structure that is under construction upon returning to it from another tabular structure. Example: navigating between matrices represented as tables while performing a manipulation.
Supporting Interaction with Domain-Specific Content
A notable challenge in providing access to domain-specific content that makes use of specialized notations lies in supporting the practice of manipulating the notation for the purpose of solving a problem. The paradigmatic example is that of manipulating equations in mathematics or the sciences.
Adequately supporting accessible manipulation and interaction is recognized in the literature as a challenge that has not been fully addressed in research carried out to date. Nevertheless, certain manipulations have been proposed, including the following.
- Copying of expressions and subexpressions (e.g., from a prior equation to the equation that is currently being edited).
- Assisted computation - automatically performing arithmetic calculations at the request of the user.
- [List to be completed.]
Music and accessibility
A more detailed summary of literature on Accessibility in the Music Domain is available.