What's the difference between MathML and Semantic MathML, especially when it comes to providing spatial structure, relations, and the various operators/symbols, etc. for STEM content? In particular, does one have more potential or benefit for screen reader (speech and/or braille) accessibility? Thanks, Kevin