Module concept_graph 

Concept Graph Builder

A semantic concept graph that extracts concepts from text, builds weighted relationships, and enables concept-based navigation and similarity search.

The graph supports:

• Automatic concept extraction via tokenization
• Co-occurrence-based edge weighting
• Explicit semantic relation tagging (synonym, hierarchical, antonym, related)
• BFS shortest-path computation
• Embedding-based or graph-topology-based similarity search
• Pruning by frequency and edge weight

Structs

CgConcept

A node in the concept graph.

CgConceptEdge

A directed (but treated as undirected for traversal) weighted edge.

CgGraphConfig

Build-time configuration.

ConceptGraphBuilder

Builds and queries a semantic concept graph.

ConceptGraphStats

Snapshot statistics about the concept graph.

ConceptId

Opaque concept index newtype.

Enums

CgConceptRelation

Semantic relation between two concepts.

Functions

cosine_similarity

Cosine similarity between two equal-length vectors.

tokenize

Split text on whitespace and ASCII punctuation; lowercase; keep tokens with at least 3 characters.