Skip to content

Next Scheduler Phases

Adrian Sampson edited this page May 13, 2016 · 2 revisions
  1. Resolve ambiguous dates from LUIS using OPAL constructs. (Iterative/nested hypothetical filtering.) Adrian is working here.
  2. Search refinement: after times are proposed, let the user say a limited vocabulary of utterances that narrow the search, like "earlier in the day." (Backtracking & incrementality.) Sarah's here.
  3. Schedule among multiple participants, or use complex preference structures. (Richer/global search strategies.)
  4. Learn individual preferences. (Reinforcement learning.)

Clone this wiki locally