RE: [voiceinteraction] review Architecture Requirements in preparation for publication

Thank you Noreen,

Yes, my concern is because of what you suspected here
> Is the concern that it is already “interpreted” because it’s output?
I am not a native speaker, but for me this does not feel correct.


Dear Debbie

Regarding
3.2.1.2 I do not see that as derived from the architecture document
Can you say a little bit more about this?
I am referring to 
• The IPA Client SHOULD forward audio data and metadata (if any) to the ASR
• The IPA Client MAY forward audio data and metadata (if any) to the ASR
To my understanding the IPA Client forwards that info to the IPA Service, which then transfers that to the ASR. So I miss one intermediate step


Dirk

From: Noreen Whysel
Sent: Wednesday, 8 March 2023 20:56
To: Deborah Dahl
Cc: Dirk Schnelle-Walka; public-voiceinteraction@w3.org
Subject: Re: [voiceinteraction] review Architecture Requirements in preparation for publication

The NLU MUST be able to interpret input Core Intent Sets


3.2.3.3 not sure if "interprets" is the right term. Core Intent Sets define the output of processing 
Not sure what a better wording would be. Do you have a suggestion?

What is the action of the interpretation of Core Intent Sets? To Parse? Recognize? Act on? Is the concern that it is already “interpreted” because it’s output?

Maps to this statement where NLU is defined in https://w3c.github.io/voiceinteraction/voice%20interaction%20drafts/paArchitecture/paArchitecture-1-3.htm#nlu

Characteristics of NLU
• The NLU is able to handle basic functionality via Core Intent Sets to enable any interaction with the user at all.
And sorry I haven’t been on the calls much lately. I’ve been so busy….

Noreen


On Mar 8, 2023, at 12:54 PM, Deborah Dahl <Dahl@conversational-technologies.com> wrote:

Thanks for your comments. I’ve updated the document with some of them, but I also have a couple of questions. 
Comments inline.
Updated draft:
https://w3c.github.io/voiceinteraction/voice%20interaction%20drafts/paRequirements/paRequirements.htm
 
From: Dirk Schnelle-Walka <dirk@switch-consulting.de> 
Sent: Friday, March 03, 2023 15:38
To: Deborah Dahl <Dahl@conversational-Technologies.com>; public-voiceinteraction@w3.org
Subject: Re: [voiceinteraction] review Architecture Requirements in preparation for publication
 
Dear Debbie,
 
Some comments from my side
 
1.7 remove "(language?)"
Done
2.1 suggestion to replace MUST by MAY, as standalone IPAs may exist
Done
3.1 similar to the previous suggestion
Did you mean that all of the points under 3.1 should be “MAY”?
 
3.1.8, 3.1.9, 3.1.10 shift to 3. before 3.1 "components..."
Done
3.1.11 remove as this is already in 3.
Done
3.2.1.2 I do not see that as derived from the architecture document
Can you say a little bit more about this?
3.2.1.3 should not be forwarded to the dialog manager but the local ASR
Done
3.2.1.4 - 3.2.1.8 similar to statement about 3.2.1.3
Modified 3.2.1.4, but 3.2.1.5-6 refer to text data, so probably shouldn’t be sent to ASR. 3.2.1.7-8 refer to multimodal data, did you want to suggest that they should also go to the ASR?
3.2.1.10 modality renderers should be replaced by client
Done
3.2.3.3 not sure if "interprets" is the right term. Core Intent Sets define the output of processing 
Not sure what a better wording would be. Do you have a suggestion?
 
Dirk
 
 
Am 23.02.2023 21:21 schrieb Deborah Dahl <Dahl@conversational-Technologies.com>:
Please review our new document, "Architecture Requirements for Intelligent Personal Assistants" located at 
https://w3c.github.io/voiceinteraction/voice%20interaction%20drafts/paRequirements/paRequirements.htm in preparation for publication 
as a Community Group Note. Please send any comments to this list. The planned publication date is March 8, unless there are any 
issues that need to be addressed prior to publication. 

Here's the document abstract: 
Abstract 
This document was prepared by reviewing version 1.3 of the Intelligent Personal Assistant Architecture Report and extracting 
requirements for the architecture that were implied by the report. This was done in order to have a standalone list of architecture 
requirements. The headings in this document correspond to the headings in the architecture report. Sections of the Architecture 
document which are not referenced here were not considered to contain any requirements. The terms "MUST", "MAY" and "SHOULD" are 
used in this document as defined in IETF RFC 2119.