Struct SemanticGraphLinker 

pub struct SemanticGraphLinker {
    pub nodes: HashMap<u64, GraphNode>,
    pub edges: Vec<SemanticEdge>,
    pub config: LinkerConfig,
}

Builds and queries a semantic similarity graph over embedding vectors.

Fields

nodes: HashMap<u64, GraphNode>

edges: Vec<SemanticEdge>

config: LinkerConfig

Implementations

impl SemanticGraphLinker

pub fn new(config: LinkerConfig) -> Self

Creates a new linker with the provided configuration.

pub fn add_node(&mut self, node: GraphNode)

Inserts a node into the graph. Any existing node with the same id is replaced (its edges are not automatically removed; call remove_node first if you want a clean replacement).

pub fn link_all(&mut self)

Links all node pairs whose cosine similarity exceeds the configured thresholds, then enforces the max_edges_per_node cap.

Calling this method more than once is safe but will duplicate edges for pairs that were already linked; it is the caller’s responsibility to clear edges first if a full rebuild is desired.

pub fn neighbors(&self, node_id: u64) -> Vec<u64>

Returns the IDs of all nodes directly adjacent to node_id.

pub fn connected_components(&self) -> Vec<Vec<u64>>

Computes connected components considering only SimilarContent and Duplicate edges (using union-find).

pub fn remove_node(&mut self, node_id: u64)

Removes a node and all edges incident to it from the graph.

pub fn stats(&self) -> GraphLinkerStats

Returns aggregate statistics for the current graph.