Compositing input

How does a multimodal application author specify what composition rules to apply?

No work released yet.

Procedural

XSLT

<xsl:if test="@emma:confidence &gt; 40">
  <xsl:copy-of select="."/>
</xsl:if>