jakarta-taglibs-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Daniel Spangler" <daniel.spang...@fluencyvoice.com>
Subject RDC: Creating a set of Nuance grammars
Date Thu, 05 May 2005 14:27:09 GMT
I have been digging through the grammars a bit and building some regular expressions to convert
them over to use the compatibility slot for Nuance and stumbled upon an issue that I wanted
to post to the list.  The issue is that according to the Nuance spec you aren't supposed to
set a slot value more than once in a grammar.  This means that doing a quick run through the
grammars and just forcing everything that uses a nameless slot to use the compatibility slot
won't be sufficient.  

The problem is that in cases like the alpha rule, the sub rules used will need to use the
nuance "return" mechanism instead.  For instance, the alpha rule uses the space rule.  The
space rule returns a nameless slot value, and the alpha rule in turn returns a nameless slot
value.  To make this compatible with nuance, we need to make the space rule use the "return"
mechanism that cannot be included as part of an interpretation result, but can be added to
a slot in the alpha rule.

This isn't a big deal to do, but means people need to be aware that the sub rules may not
work as expected if used directly within an RDC.  For example, if the space rule uses the
"return" mechanism where no interpretation results are present, and someone builds an RDC
that accesses the space rule directly; it won't behave as expected when the nuance grammar
is used.

Daniel

 

 

  




 

________________________________________________________________________
This email has been scanned for all viruses by Netscalibur Mail Scanner, powered by MessageLabs.
For more information on a proactive email security service working around the clock, around
the globe, visit
http://www.netscalibur.co.uk/scanner/index.html
________________________________________________________________________
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message