Struct VectorQueryExpander 

pub struct VectorQueryExpander { /* private fields */ }

Vector-based semantic query expander.

Maintains a synonym registry where each term maps to a list of (synonym_string, embedding_vector) pairs. Given a query term and its embedding, the expander finds synonyms whose cosine similarity exceeds VectorExpanderConfig::similarity_threshold, weights them by similarity × decay^rank, and returns a combined (L2-normalised) vector.

Implementations

impl VectorQueryExpander

pub fn new(config: VectorExpanderConfig) -> Self

Create a new expander with the given configuration and an empty synonym registry.

pub fn add_synonym(&mut self, term: &str, synonym: &str, vector: Vec<f64>)

Register a synonym for term together with its embedding vector.

pub fn expand_query( &mut self, query_term: &str, query_vector: &[f64], ) -> VectorExpandedQuery

Expand a query by finding matching synonyms, weighting them, and producing a combined vector.

Steps:

1. Look up synonyms for query_term.
2. Compute cosine similarity of each synonym vector to query_vector.
3. Filter by similarity_threshold.
4. Sort descending by similarity.
5. Take at most max_expansions.
6. Weight each expansion by similarity × decay^rank (rank starts at 0).
7. Combine vectors via weighted average and L2-normalise.

pub fn combine_vectors( original: &[f64], expansions: &[(Vec<f64>, f64)], ) -> Vec<f64>

Weighted average of original and expansions, then L2-normalise.

Formula: (original + Σ(weight_i × expansion_i)) / (1 + Σ weight_i)

pub fn cosine_similarity(a: &[f64], b: &[f64]) -> f64

Cosine similarity between two vectors.

Returns 0.0 when either vector has zero magnitude.

pub fn remove_synonym(&mut self, term: &str, synonym: &str) -> bool

Remove a specific synonym for term. Returns true if found and removed.

pub fn synonym_count(&self, term: &str) -> usize

Number of synonyms registered for term (0 if unknown).

pub fn clear_synonyms(&mut self)

Remove all synonym entries.

pub fn stats(&self) -> VectorExpanderStats

Return a snapshot of the current runtime statistics.