This document was generated from CDN thread
Created by: null on 26-01-2006 03:06:32 PM
Hi,
after going through a first evaluation, it is not clear how Audium can implement a mixed initiative speech dialog, in which the user can take the initiative in a dialog to a greater or lesser degree.
In a mixed initiative state, the application starts with a prompt that attempts to elicit multiple pieces of related information. A typical example is a reservation system where a prompt asks for flight details and the application expects something like â¿¿I want to fly from Boston to Philadelphia on May fifth.â¿¿ Obvious variables are â¿¿fromCity, toCity, and flightDate.â¿¿ So far this is a â¿¿directedâ¿¿ dialogue. What makes this â¿¿mixed initiativeâ¿¿ is that it might be constructed to accept just parts of the response (e.g., â¿¿I want to fly from Boston to Philadelphiaâ¿¿); then the application prompts again for the missing details (e.g., When do you want to leave?). VoiceXML's Form Interpretation Algorithm supports this feature.
How can this type of dialog be built in Audium? Is it possible with the existing set of Audium elements?
Regards,
Dirk
Subject: RE: Creating Mixed Initiative dialogs.
Replied by: Michael Bochynski on 30-01-2006 04:44:03 PM
Hi Dirk,
Audium provides the mixied initiative functionality, and the developer does not have to resort to programming. It is certainly possible with the existing set of AUdium elements.
To implement mixed initiative dialogs, a Form element is used to capture multiple slots, and then a Decision element directly follows it to determine if any slots were not filled. The Decision element then branches the call flow to acquire the missing parts.
We have a sample application that demonstrates this behavior. The caller is asked which city and state they would like to hear the weather for, and if the caller does not provide enough information, they are prompted only for the missing parts before the weather forecast is played to them. The valid city and state pairs are "San Francisco, California", "Los Angeles, California", and "Boston, Massachusetts". You can find the sample app attached to this post.
The application has been tested to work on SpeechWorks OSB 1.2 with OSR 1.0/1.1. Open the application in Studio and deploy it if you would like to call in to test it. However, there are some additional steps that must be completed before you can call in:
1) There are 4 URLs that must be updated. These include: 1 grammar URI in "CityStateCapture", 1 grammar URI in "CityCapture", 1 Audio URI in "WeatherAudio" (Initial, audio item 3) and 1 Audio URI in "WeatherAudio2" (Initial, audio item 3).
2) Two external grammar files are referenced by the application. They are located in a folder named "grammar" in the deploy/data/misc/ directory. Before starting the application server, copy the entire grammar folder to the test machine's application server ROOT directory.
3) Three .vox audio files are located in the "audio" folder under: deploy/data/misc/. The entire audio folder must be copied to the test machine's application server ROOT directory before starting the application.
Here's a sample transcript of the dialog between a novice user and the system:
C: Welcome to the weather information service. Buy Joe's Spicy Shrimp Sauce.
C: For what city and state would you like the weather?
H: Uh, California.
C: Please say the city in for which you want the weather.
H: San Francisco, please.
C: Do you want to hear the weather for [San Francisco, California]?
H: No.
C: For what city and state would you like the weather?
H: Los Angeles.
C: Do you want to hear the weather for [Los Angeles, California]?
H: Yes
C: Don't forget, buy Joe's Spicy Shrimp Sauce tonight!
C: Mostly sunny today with highs in the 80s. Lows tonight from the low 60s ...
However, the dialog for the user familiar with prompts would be:
C: Welcome to the weather information service. Buy Joe's Spicy Shrimp Sauce.
C: For what city and state would you like the weather?
H: Boston, Massachusetts.
C: Don't forget, buy Joe's Spicy Shrimp Sauce tonight!
C: Mostly sunny today with highs in the 70s. Lows tonight from the low 50s.
I hope that you will find this explanation and sample app useful. Please let us know if you have any further questions.
Regards,
Michael
Subject: RE: Creating Mixed Initiative dialogs.
Replied by: null on 31-01-2006 10:39:12 AM
Hi Michael,
You refer to an attached sample application - "You can find the sample app attached to this post." How can I get it?
Regards,
Dirk
Subject: RE: Creating Mixed Initiative dialogs.
Replied by: Michael Bochynski on 31-01-2006 02:59:29 PM
Dirk,
You should be able to see the attachment to the previous post. It was a permissions issue.
Regards,
Michael
