Documentation
¶
Overview ¶
Package probability provides abstractions for how to compute how precision and recall is affected by measurements.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type MeSHExplosionsRatio ¶
type MeSHExplosionsRatio struct{}
MeSHExplosionsRatio measures and predicts how mesh explosions affect retrieval performance.
func (MeSHExplosionsRatio) ComputeAdditionProbability ¶
func (MeSHExplosionsRatio) ComputeAdditionProbability(m float64) PredictionPair
ComputeAdditionProbability NOT IMPLEMENTED.
func (MeSHExplosionsRatio) ComputeReductionProbability ¶
func (MeSHExplosionsRatio) ComputeReductionProbability(m float64) PredictionPair
ComputeReductionProbability NOT IMPLEMENTED.
func (MeSHExplosionsRatio) Execute ¶
func (MeSHExplosionsRatio) Execute(pq pipeline.Query, s stats.StatisticsSource) (float64, error)
Execute returns the ratio of exploded MeSH queries to the number of total MeSH queries in the search strategy.
type PredictionPair ¶
PredictionPair is a pair of probabilities from predicting how a Measurement affects retrieval performance.
func NewPredictionPair ¶
func NewPredictionPair(precision, recall float64) PredictionPair
NewPredictionPair creates a new prediction pair.
type ProbabilisticMeasurement ¶
type ProbabilisticMeasurement struct {
analysis.Measurement
Probability
}
ProbabilisticMeasurement is the union of a Measurement (i.e. some QPP), and a probability (i.e. how this Measurement affects retrieval performance).
func NewProbabilisticMeasurement ¶
func NewProbabilisticMeasurement(measurement Probability) ProbabilisticMeasurement
NewProbabilisticMeasurement creates a ProbabilisticMeasurement from a single Probability implementation. This should ensure at compile-time that a ProbabilisticMeasurement has both interfaces implemented.
func (ProbabilisticMeasurement) Compute ¶
func (pm ProbabilisticMeasurement) Compute(q pipeline.Query, s stats.StatisticsSource) (ProbabilisticMeasurementResult, error)
Compute a Measurement and how that Measurement affects precision and recall.
type ProbabilisticMeasurementResult ¶
type ProbabilisticMeasurementResult struct {
Measurement float64
Addition PredictionPair
Reduction PredictionPair
}
ProbabilisticMeasurementResult is the result of computing the likelihoods for how a Measurement affects precision and recall by adding and removing to a query.
type Probability ¶
type Probability interface {
ComputeAdditionProbability(float64) PredictionPair
ComputeReductionProbability(float64) PredictionPair
}
Probability computes the likelihood of a Measurement affecting precision and recall. For this interface to be properly implemented, four probabilities must be implemented: how the Measurement affects retrieval performance if the query is rewritten to add or remove aspects that increase or decrease the Measurement.
Source Files