Enum UncertainPrediction 

pub enum UncertainPrediction {
    Single {
        entity_type: EntityType,
        confidence: f64,
    },
    Distributed(TypeDistribution),
    Abstained(Abstention),
}

A prediction that may include uncertainty or abstention.

This is the core type for selective prediction, replacing simple (EntityType, confidence) pairs with richer uncertainty information.

Variants

• Single: A single type prediction with confidence (traditional)
• Distributed: Probabilities over multiple types (soft prediction)
• Abstained: Explicit refusal to predict with reason

Variants

Single

Single type prediction (traditional NER output).

Fields

entity_type: EntityType

The predicted type

confidence: f64

Confidence score [0, 1]

Distributed(TypeDistribution)

Distribution over types (soft prediction).

Abstained(Abstention)

Explicit abstention with reason.

Implementations

impl UncertainPrediction

pub fn from_type(entity_type: EntityType, confidence: f64) -> Self

Create a single-type prediction.

pub fn distributed(dist: TypeDistribution) -> Self

Create a distributed prediction.

pub fn abstain(reason: Abstention) -> Self

Create an abstention.

pub fn abstain_low_confidence(max_score: f64) -> Self

Create abstention due to low confidence.

pub fn abstain_ambiguous(top_types: Vec<EntityType>, margin: f64) -> Self

Create abstention due to ambiguity.

pub fn is_abstention(&self) -> bool

Check if this is an abstention.

pub fn is_confident(&self, threshold: f64) -> bool

Check if this prediction is confident (above threshold).

pub fn best(&self) -> Option<(&EntityType, f64)>

Get the best prediction type and confidence.

Returns None if abstained or distribution is empty.

pub fn get_type(&self) -> Option<&EntityType>

Get the entity type if prediction is confident.

Returns None if abstained, distributed, or below threshold.

pub fn confidence(&self) -> f64

Get confidence score.

Returns 0.0 for abstentions, max prob for distributions.

pub fn distribution(&self) -> Option<&TypeDistribution>

Get the type distribution if available.

pub fn abstention_reason(&self) -> Option<&Abstention>

Get the abstention reason if abstained.

pub fn with_threshold(self, threshold: f64) -> Self

Convert to a single prediction, applying threshold for abstention.

If the prediction is below threshold, converts to abstention.

Trait Implementations

impl Clone for UncertainPrediction

fn clone(&self) -> UncertainPrediction

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for UncertainPrediction

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for UncertainPrediction

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for UncertainPrediction

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for UncertainPrediction

fn eq(&self, other: &UncertainPrediction) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for UncertainPrediction

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for UncertainPrediction