Struct SemanticEntityLinker 

pub struct SemanticEntityLinker {
    pub entities: HashMap<u64, KbEntity>,
    pub name_index: HashMap<String, u64>,
    pub alias_index: HashMap<String, u64>,
    pub config: LinkerConfig,
    pub stats: LinkerStats,
}

A linker that resolves mention texts to canonical knowledge-base entities.

Resolution is attempted in priority order:

1. Exact name match (lowercase comparison against canonical names)
2. Alias match (lowercase comparison against known aliases)
3. Embedding similarity (cosine similarity above configured threshold)

Fields

entities: HashMap<u64, KbEntity>

All registered entities, keyed by entity_id.

name_index: HashMap<String, u64>

Lowercase canonical name → entity_id.

alias_index: HashMap<String, u64>

Lowercase alias → entity_id.

config: LinkerConfig

Configuration parameters.

stats: LinkerStats

Running statistics.

Implementations

impl SemanticEntityLinker

pub fn new(config: LinkerConfig) -> Self

Create a new, empty linker with the given configuration.

pub fn register(&mut self, entity: KbEntity)

Register an entity in the knowledge base.

Indexes the entity’s canonical name and all aliases (lowercased). If an entity with the same entity_id was already registered, it is replaced.

pub fn link( &mut self, mention_text: &str, mention_embedding: Option<&[f32]>, ) -> Option<LinkedMention>

Attempt to link a single mention text to an entity.

Resolution priority:

1. ExactMatch — lowercase mention in name_index
2. Alias — lowercase mention in alias_index
3. EmbeddingMatch — highest cosine similarity ≥ embedding_threshold

Returns None and increments unlinked_count if no match is found.

pub fn link_batch( &mut self, mentions: Vec<(String, Option<Vec<f32>>)>, ) -> Vec<Option<LinkedMention>>

Link a batch of mentions, each paired with an optional embedding.

Returns one Option<LinkedMention> per input mention, in the same order.

pub fn entity(&self, entity_id: u64) -> Option<&KbEntity>

Look up a registered entity by its entity_id.

pub fn stats(&self) -> &LinkerStats

Return a reference to the current linker statistics.