- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Tue, 4 Aug 2009 15:58:28 +0100
- To: W3C OWL Working Group <public-owl-wg@w3.org>
Structural Spec. METHODOLOGY: I tried to read the document using VoiceOver and inspected the rendering of the document by Fangs: http://sourceforge.net/projects/fangs/ Fangs is by far the least tedious I also tried some accessibility checkers, e.g.,: http://www.etre.com/tools/accessibilitycheck/ (0 "must fix", 1 "should fix" "No link is provided for downloading the PDF plug-in", but that seems bogus to me, and 1 "may fix" "a tags need to be correctly nested and closed"...no idea where they are but I presume we can catch them with validation). Many of the online checkers time out with our documents :) EXECUTIVE SUMMARY: It isn't bad at all. I seem some tweaks that could improve it and some hard problems that aren't obviously solvable. The biggest wins in my opinion: 1) Make the alt text of all UML diagrams "UML diagram" 2) Make the layout tables for the examples better for screen reading (perhaps with summary or hidden headers) and inline "example". 3) Port the examples to manchester syntax. I don't know how to make the punctuation symbols and BNF accessible with out a rewrite. MY OBSERVATIONS I see no skip to main/skip to TOC. There's a *lot* of boilerplate up front that, well, sucks to wade through. "Hide FSS in Example" <-- "FSS" needs to be read out. <div class="anexample"> <-- The generated text from the CSS isn't read aloud. ""ObjectUnionOf left paren a colon Person a colon Animal right paren"" Punctuation problems. By default, loads of punctuation is not read aloud. We use punctuation for significant stuff. Not sure what to do with that. This problem shows up in the BNF since in some readers, square brackets aren't read. But in general: left brace ClassExpression right brace doesn't seem all that intuitive. (Plus the use of bold, etc.) I'm not sure what to do with that. I could rewrite everything...but...but...I really don't want to! Unicode U+digets is helpful on the one hand (for non read symbols) but seems kinda horrible. The obvious thing would be to use the standard names...but omg. Table 3 may need a summary to indicate that the columns have no meaning. Better yet would be to group the OWL terms, rdf terms, xsd terms and perhaps use lists. This is pretty much a layout table as it stands. """<img alt="The Structure of OWL 2 Ontologies" border="0" height="274" src="Ontology.gif" width="617"/><br/> <span class="caption">Figure 1.</span> The Structure of OWL 2 Ontologies </p>""" This gets "The Structure of OWL 2 Ontologies" repeated twice. E.g., Fangs: """GraphicThe Structure of OWL two Ontologies Figure one . The Structure of OWL two Ontologies""" I'd suggest "UML diagram" as the alt text which would read as: """GraphicUML diagram Figure one . The Structure of OWL two Ontologies""" 3.6, the algorithm is laid out with a table. Either it should have headers (step label, step) or a at least a summary: "Table with two columns and twelve rows CP one Make AllDoc and Processed equal to the empty set, and make ToProcess equal to the set containing only the IRI GI . " The hiding might not be strong enough: """Table with two columns and one row AnnotationAssertion left paren rdfs colon comment a colon Peter quote The father of the Griffin family from Quahog. quote right paren This axiom provides a comment for the IRI a colon Peter . Table endTable caption colon RDF colon Table with two columns and one row""" As you can see the second (hidden) table is read out, but as empty. Plus these are layout tables. I'm not exactly sure what to do with this other than live with it. A summary might help (explaining that the first bit is the syntax the second bit the explanation). I'm unclear that I would want functional syntax or RDF read to me. The obvious other candidate is Manchester syntax which we know at least one blind reader would prefer. In general, I expect that the examples would be the most intelligible and valuable thing this document offers, along with a good deal of the explanatory text. I expect one could reasonably become familiar with the languages with the headings and example alone. (and it is the referenceynature of the document). Cheers, Bijan.
Received on Tuesday, 4 August 2009 14:54:06 UTC