Struct EntityLinker 

pub struct EntityLinker {
    pub min_mention_len: usize,
    pub threshold: f64,
    /* private fields */
}

String-to-entity linker backed by an in-memory knowledge base.

Fields

min_mention_len: usize

Minimum character length of a text span to be considered as a mention.

threshold: f64

Minimum combined score for a candidate to be returned.

Implementations

impl EntityLinker

pub fn new(threshold: f64) -> Self

Create an empty linker with the given minimum combined score threshold.

pub fn add_entity(&mut self, entity: Entity)

Add a single entity to the knowledge base and update the index.

pub fn add_entities(&mut self, entities: Vec<Entity>)

Add multiple entities in bulk and rebuild the index.

pub fn entity_count(&self) -> usize

Total number of entities in the knowledge base.

pub fn detect_mentions(&self, text: &str) -> Vec<EntityMention>

Scan text for all known entity labels / aliases and return mentions.

Both exact (case-insensitive) and fuzzy (similarity ≥ 0.7) matches are considered. Overlapping spans from the same starting position are de-duplicated in favour of the longer match.

pub fn link(&self, text: &str) -> Result<Vec<LinkedEntity>, LinkerError>

Link all mentions found in text to their best entity candidates.

Errors

• LinkerError::EmptyText if text is empty after trimming.
• LinkerError::EmptyKnowledgeBase if no entities have been added.

pub fn link_mention(&self, mention: &str) -> Vec<LinkCandidate>

Find all entity candidates for an arbitrary mention string.

Searches all entity labels and aliases using both exact (lowercase) and fuzzy matching. Results are sorted by score descending.

pub fn find_by_iri(&self, iri: &str) -> Option<&Entity>

Look up an entity by its exact IRI.

pub fn string_similarity(a: &str, b: &str) -> f64

Normalised edit-distance similarity: 1 − edit_distance(a, b) / max(|a|, |b|).

Returns 1.0 for identical strings and 0.0 when the edit distance equals the length of the longer string.