Expand description
What an item expects of each understanding task, beside what it expects of the turn.
A turn-level expectation says what happened; these say which task got it right or wrong: how the message split into units, which operation and record each request reached, and which arguments were given or not. They are parsed and validated with the rest of the item, and scored against the task records a turn leaves behind.
[[expect.understanding.units]]
kind = "request"
words = "the subject needs changing"
operation = "trip.set_name"
record = "trip-1"
arguments = { value = { not_given = true } }Structs§
- Argument
Expectation - What one argument must come out as: a value, or explicitly not given.
- Task
Scores - How each understanding task did on one sample:
Nonewhere the item states nothing that task decides. - Understanding
Expectation - The units a message splits into, in message order.
- Unit
Expectation - One unit of a message, and what the tasks after segmentation must make of it.
Enums§
- Unit
Kind - The kinds of unit segmentation distinguishes.