Struct IntentClassifier 

pub struct IntentClassifier {
    pub intents: Vec<String>,
    pub patterns: Vec<Vec<String>>,
}

Pattern-matching intent classifier.

Each intent is represented by a name and a list of keyword patterns. An utterance is matched by counting how many patterns contain at least one word from the utterance; the intent with the most matches wins.

Example

use scirs2_text::dialog::{IntentClassifier, classify_intent};

let mut clf = IntentClassifier::new();
clf.add_intent("book_flight", vec!["book", "flight", "fly", "ticket"]);
clf.add_intent("cancel", vec!["cancel", "undo", "remove"]);

let (intent, confidence) = classify_intent("I want to book a flight", &clf);
assert_eq!(intent, "book_flight");
assert!(confidence > 0.0);

Fields

intents: Vec<String>

Registered intent names in registration order.

patterns: Vec<Vec<String>>

Patterns for each intent (parallel to intents).

Each element is a list of keyword strings that evidence the intent.

Implementations

impl IntentClassifier

pub fn new() -> Self

Create an empty classifier.

pub fn add_intent(&mut self, name: &str, patterns: Vec<&str>)

Register a new intent with its keyword patterns.

All pattern strings are lower-cased and stored as-is; matching is performed case-insensitively.

pub fn len(&self) -> usize

Return the number of registered intents.

pub fn is_empty(&self) -> bool

Return true when no intents are registered.

Trait Implementations

impl Clone for IntentClassifier

fn clone(&self) -> IntentClassifier

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for IntentClassifier

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

Formats the value using the given formatter.

impl Default for IntentClassifier

fn default() -> IntentClassifier

Returns the “default value” for a type.