Struct GraphEmbedder 

pub struct GraphEmbedder;

Graph embedding engine: Node2Vec-inspired random walks and structural embeddings

Implementations

impl GraphEmbedder

pub fn random_walks(graph: &Graph, config: &WalkConfig) -> Vec<Vec<usize>>

Generate biased random walks for all nodes.

Uses Node2Vec-style second-order biased walks governed by p and q. When p=1 and q=1 the walk degenerates to a uniform random walk (DeepWalk).

Returns a list of walks, each being a sequence of node IDs.

pub fn embed(graph: &Graph, config: &WalkConfig, dim: usize) -> EmbeddingResult

Generate random-walk-based embeddings.

Each node’s embedding is derived from its walk co-occurrence profile (simplified: averaged node-ID features from walk context).

pub fn structural_embedding(graph: &Graph, dim: usize) -> Vec<NodeEmbedding>

Compute structural embeddings based purely on local neighborhood topology.

Each node’s embedding aggregates neighbour degree statistics projected into dim.

pub fn node_similarity(a: &NodeEmbedding, b: &NodeEmbedding) -> f32

Cosine similarity between two node embeddings, in range [-1, 1].

pub fn neighbors(graph: &Graph, node: usize) -> Vec<usize>

Return all neighbors of node (both directions for undirected usage).

pub fn degree(graph: &Graph, node: usize) -> usize

Degree of a node (number of unique neighbors).