Struct TripleExtractor 

pub struct TripleExtractor { /* private fields */ }

Extracts subject–predicate–object triples from free text

Implementations

impl TripleExtractor

pub fn new(config: ExtractionConfig) -> Self

Create an extractor with no built-in patterns (only the supplied config).

pub fn with_defaults(config: ExtractionConfig) -> Self

Create an extractor pre-loaded with common English relation patterns.

Built-in patterns cover: is, has, works at, located in, founded by, created by, known as, born in, part of.

pub fn add_pattern(&mut self, pattern: ExtractionPattern)

Add a custom pattern to the extractor.

pub fn pattern_count(&self) -> usize

Returns the number of registered patterns.

pub fn extract(&self, text: &str) -> Vec<TextTriple>

Extract triples from text, splitting on sentence boundaries ('.', '!', '?').

pub fn extract_sentence(&self, sentence: &str) -> Vec<TextTriple>

Extract triples from a single sentence string.

pub fn normalize_predicate(predicate: &str) -> String

Normalise a raw predicate string: lowercase, remove leading articles/stop words, collapse consecutive spaces, and trim.

pub fn confidence_for_pattern( matched_words: usize, total_pattern_words: usize, ) -> f64

Confidence based on how many predicate words were matched.

Returns matched_words / total_pattern_words, clamped to [0.0, 1.0].

pub fn to_knowledge_graph( triples: &[TextTriple], ) -> Vec<(String, String, String)>

Convert a slice of TextTriples to plain (subject, predicate, object) tuples.