[Chandler-dev] Natural language parsing in Chandler

[Philippe Bossut]

Ted Leung wrote:
> The whole command line approach seems wrong to me.  Agenda, and 
> Apple's Newton, both of which ran on significantly less capable 
> hardware than we have today, were able to do recognition of people, 
> dates, etc without the need for a semantic hint such as /event or 
> /task.   It might be that the first cut of recognition needs those 
> hints to be implementable, but I think that a longer range goal should 
> to reduce the need for these kinds of hints.
I haven't used any of those products so I can't tell. The problem I see 
though is that, if we have one only text field, we need to disambiguate 
the intent of the user so that we know if the user is searching or 
creating something.

e.g. I type "tomorrow morning", it could be "create an event with the 
start date set to tomorrow morning" or "search and display all events 
(or tasks with ticklers) happening tomorrow morning". I don't see how 
you can disambiguate the intent without having some parameter entered by 
the user (in the form of a /create or /search command) or 2 fields 
(which adds extra clutter to the UI).

I liked Jeffrey's (and others) proposal though that was a mix of both: a 
unique field that would display the current "intent" using an icon at 
the beginning of the field (you can click the icon and select another 
intent/command if you want) and allow to flip the intent using a command 
introduced by "/" (so you don't have to click anything for advanced users).

What you say however and that makes sense to me is that we can restrict 
the vocabulary of commands and let the recognizer create the appropriate 
item, i.e. having a unique /create command for all items (and let the 
NLP do the choice of item to create, stamp, etc...) instead of one 
command per kind.

Cheers,
- Philippe