Struct DeepWalk 

pub struct DeepWalk<N: Node> { /* private fields */ }

DeepWalk embedding algorithm

Learns node embeddings using uniform random walks followed by skip-gram optimization. Supports both negative sampling and hierarchical softmax.

Implementations

impl<N: Node> DeepWalk<N>

pub fn new(config: DeepWalkConfig) -> Self

Create a new DeepWalk instance with negative sampling (default)

pub fn with_hierarchical_softmax(config: DeepWalkConfig) -> Self

Create a new DeepWalk instance with hierarchical softmax

pub fn set_mode(&mut self, mode: DeepWalkMode)

Set the training mode

pub fn mode(&self) -> DeepWalkMode

Get the current training mode

pub fn generate_walks<E, Ix>( &mut self, graph: &Graph<N, E, Ix>, ) -> Result<Vec<RandomWalk<N>>>

where N: Clone + Debug, E: EdgeWeight, Ix: IndexType,

Generate training data (uniform random walks) on undirected graph

pub fn generate_walks_digraph<E, Ix>( &mut self, graph: &DiGraph<N, E, Ix>, ) -> Result<Vec<RandomWalk<N>>>

where N: Clone + Debug, E: EdgeWeight, Ix: IndexType,

Generate training data (uniform random walks) on directed graph

pub fn train<E, Ix>(&mut self, graph: &Graph<N, E, Ix>) -> Result<()>

where N: Clone + Debug, E: EdgeWeight, Ix: IndexType,

Train the DeepWalk model on an undirected graph

pub fn train_digraph<E, Ix>(&mut self, graph: &DiGraph<N, E, Ix>) -> Result<()>

where N: Clone + Debug, E: EdgeWeight, Ix: IndexType,

Train the DeepWalk model on a directed graph

pub fn model(&self) -> &EmbeddingModel<N>

Get the trained model

pub fn model_mut(&mut self) -> &mut EmbeddingModel<N>

Get mutable reference to the model