Struct SemanticEngine 

pub struct SemanticEngine {
    pub ngram: CharNgramIndex,
    pub cooccurrence: CoOccurrenceIndex,
    pub synonym: SynonymExpander,
    pub weights: SemanticWeights,
}

Combined semantic matching: 0.4ngram + 0.4cooccurrence + 0.2synonym. Two-phase query_fast: phase 1 ngram+synonym (0.6/0.4), phase 2 re-rank top 3K with cooccurrence. Replaces: semantic_v2.py SemanticEngine

Fields

ngram: CharNgramIndex

cooccurrence: CoOccurrenceIndex

synonym: SynonymExpander

weights: SemanticWeights

Implementations

impl SemanticEngine

pub fn build(graph: &Graph, weights: SemanticWeights) -> M1ndResult<Self>

Build all three indexes from the graph. Replaces: semantic_v2.py SemanticEngine.init()

pub fn query( &self, graph: &Graph, query: &str, top_k: usize, ) -> M1ndResult<Vec<(NodeId, FiniteF32)>>

Full query: score all nodes, return top_k. Weight: ngram0.4 + cooccurrence0.4 + synonym*0.2. Replaces: semantic_v2.py SemanticEngine.query()

pub fn query_fast( &self, graph: &Graph, query: &str, top_k: usize, ) -> M1ndResult<Vec<(NodeId, FiniteF32)>>

Fast two-phase query. Phase 1: ngram+synonym (0.6/0.4) -> candidates (3*top_k). Phase 2: re-rank candidates with cooccurrence. Replaces: semantic_v2.py SemanticEngine.query_fast()