ECIG processing rules

[Jacob Westfall <jake@jpw.biz>]

Hi,
	As per the Profiles SC call yesterday, I am providing an example (more if necessary) of where the ECIG processing rules make syncing between an EAS text crawl/closed captioning derived from CAP elements and a human recorded EAS message difficult.  Of course EAS broadcasts today aren't usually in sync anyways... but that is presumably part of the improvement CAP is supposed to provide.  
	I think the overriding concern is ensuring that users create "good" CAP messages first and foremost that utilize all of the CAP elements as intended.  The concern with any sort of processing rules, ECIG or otherwise, are that bias will creep in on the originator side and the data in the elements will be constructed to conform to those rules rather than the intent of the CAP elements.

Starting with a very simple example from the CAP specification, which I hope is a good example of a CAP message :)

"The National Weather Service has issued a Severe Thunderstorm Warning for all of Tuolumne County, all of Calaveras County, all of Alpine County, beginning at 2:57 PM and ending at 4 PM.  This is the NATIONAL WEATHER SERVICE SACRAMENTO CA.  SEVERE THUNDERSTORM WARNING.  SEVERE THUNDERSTORM.  EXTREME NORTH CENTRAL TUOLUMNE COUNTY IN CALIFORNIA, EXTREME NORTHEASTERN CALAVERAS COUNTY IN CALIFORNIA, SOUTHWESTERN ALPINE COUNTY IN CALIFORNIA.  AT 254 PM PDT...NATIONAL WEATHER SERVICE DOPPLER RADAR INDICATED A SEVERE THUNDERSTORM OVER SOUTH CENTRAL ALPINE COUNTY...OR ABOUT 18 MILES SOUTHEAST OF KIRKWOOD...MOVING SOUTHWEST AT 5 MPH. HAIL...INTENSE RAIN AND STRONG DAMAGING WINDS ARE LIKELY WITH THIS STORM.  TAKE COVER IN A SUBSTANTIAL SHELTER UNTIL THE STORM PASSES."

As you can see the message becomes very repetitive in the middle portion.  This is especially a problem if a proper headline is created.  In this case the headline is "Severe Thunderstorm Warning" but a better version would actually be the initial sentence itself.  While a text-to-speech system doesn't care, a human speaker may look at this script and notice this and want to re-work the EAS script.  Because the initial sentence generated using the EAS Header Code, which in turn used the CAP elements, already contains the who, what, where, and when, the next part of importance is the Instruction.  So a reworked version might look like,

"The National Weather Service has issued a Severe Thunderstorm Warning for all of Tuolumne County, all of Calaveras County, all of Alpine County, beginning at 2:57 PM and ending at 4 PM. TAKE COVER IN A SUBSTANTIAL SHELTER UNTIL THE STORM PASSES.  AT 254 PM PDT...NATIONAL WEATHER SERVICE DOPPLER RADAR INDICATED A SEVERE THUNDERSTORM OVER SOUTH CENTRAL ALPINE COUNTY...OR ABOUT 18 MILES SOUTHEAST OF KIRKWOOD...MOVING SOUTHWEST AT 5 MPH. HAIL...INTENSE RAIN AND STRONG DAMAGING WINDS ARE LIKELY WITH THIS STORM."

However if the human speaker does rework the recorded audio script, it no longer matches the text version.  And it gets even worse for things like Amber alerts and Hazmat which are less defined than weather incidents.  I think the key problem with the current ECIG processing rules is the fact that the initial sentence that is generated by the header values needs to be properly incorporated into the entire set of rules and not stand alone.

But the key question underlying this whole concern, probably for the IPAWS folks, would be whether there is a Profile "requirement", perhaps because of the Americans with Disabilities Act, that the EAS recorded audio match the on-screen EAS text "exactly".

-- 
jake@jpw.biz
--