BinRoot
commented
5 years ago will those developing apps on mobile devices also be included, or being W3C is that out of scope. My feeling is that mobile apps is where it's most needed.
I see no problem including mobile apps, since the underlying conversational interface still relies heavily on the web, and the mobile device may act as the front-end.
"Authoring and debugging of multi-modal conversational interfaces" - that sound like you are getting into the NLU tool space, which I think the standard should be independent of. UI guidelines fine, but not authoring tools.
Right, the standardization of debugging and running chatbots should try to steer away from any specific AI/ML/NLU technology. Persistent storage/representation of dialogue flow, for sharing and editing, could be decoupled from the underlying engine that runs the bot; for example, consider exporting an Alexa skill into a representation (DMPL) that can then be imported into the Google Home, or vice versa.
scoring chatbot conversations for systems development and comparison
Neat, I added it to the "To Do" list: https://conversational-interfaces.github.io/ I think it's in scope, since it can be framed as a utility model.
so that a single bot can be accessible from multiple devices
A standard API for bot-to-UI is indeed useful. Added to the "To Do" list.
federation of bots
This is a brilliant idea, added to the "To Do" list. I believe the utility model work can be relevant here: each bot/skill comes with a standard description of what it can do, defined by the utility it creates.
Just from a quick reading:
"web developers" - will those developing apps on mobile devices also be included, or being W3C is that out of scope. My feeling is that mobile apps is where it's most needed.
"Authoring and debugging of multi-modal conversational interfaces" - that sound like you are getting into the NLU tool space, which I think the standard should be independent of. UI guidelines fine, but not authoring tools.
"Visualization techniques of dialogue flow" - that would be very useful
"Data representation of real-world dialogue for storage and learning" - one thing I'd love to extend that with is a common approach to scoring chatbot conversations for systems development and comparison. It will alsoways be relatively subjective, but we've foudn a system based around Grice's Conversational Maxims (https://www.sas.upenn.edu/~haroldfs/dravling/grice.html) to work well.
Dialogue management programming language (DMPL) - not sure I see how this is going to help, looks like an extra layer that needs to be implemented by/adopted by developers, rather than helping with the user experience or the interoperability of systems.
What I'd rather see is: a) a standard on what the "bot" exchanges with the user interface, so that a single bot can be accessible from multiple devices (perhaps a bit like WebVR) . Don't know what there is already on this but we have an internal standard we use to specify conversational features (e.g. text, images, buttons, sliders, stars etc) by the bot (plus flags for additional/debugging into) with realisation left to the end device. Long way from a standard but happy to share. b) a standard on inter-bot communication so one bot can find out the scope of other bots, choose one to work with and then pass through user requests and handle responses back to the user so that you get a federation of bot.
It may be that this group is not looking at what I thought was meant by conversational interfaces, if so apologies.