edu.umass.cs.automan.core.policy.aggregation
Determines whether a policy allows for canceling running tasks.
Determines whether a policy allows for canceling running tasks. If true, is_done will be called more often, increasing the required confidence level to terminate. Note that an early termination check is always conducted when timeouts occur regardless of this setting.
Returns a list of blacklisted worker_ids given a set of tasks, completed or not.
Returns a list of blacklisted worker_ids given a set of tasks, completed or not.
The complete list of tasks.
A list of worker IDs.
Returns true if the strategy has enough data to stop scheduling work.
Returns true if the strategy has enough data to stop scheduling work.
The complete list of scheduled tasks.
The number of times this function has been called, inclusive.
true iff done.
Calculate the number of new tasks to schedule.
Calculate the number of new tasks to schedule.
Partitions a set of tasks into those that should be marked as duplicate and those that should not be.
Partitions a set of tasks into those that should be marked as duplicate and those that should not be. All tasks passed in are passed back out.
A list of tasks
(list of non-duplicate tasks, list of duplicate tasks)
Returns a string explaining why the worker's answer was not accepted.
Returns a string explaining why the worker's answer was not accepted.
The list of accepted tasks. Used to determine the correct answer.
Explanation string.
Returns the top answer.
Returns the top answer.
The complete list of tasks.
The number of times is_done has been called.
Top answer
Returns an appropriate response for when the computation ran out of money.
Returns an appropriate response for when the computation ran out of money.
The complete list of tasks.
The smallest amount of money needed to complete the computation under optimistic assumptions.
The amount of money we have.
The number of times is_done has been called.
A low-confidence or over-budget answer.
Computes the number of tasks needed to satisfy the quality-control algorithm given the already-collected list of tasks.
Computes the number of tasks needed to satisfy the quality-control algorithm given the already-collected list of tasks. Returns only newly-created tasks.
The complete list of previously-scheduled tasks
True if any of the latest batch of tasks suffered a timeout.
A list of new tasks to schedule on the backend.