Struct GraphSage 

pub struct GraphSage { /* private fields */ }

GraphSAGE model for inductive node embedding

Implements the GraphSAGE algorithm from Hamilton et al. (2017). Key property: can generate embeddings for nodes not seen during training by aggregating from their neighborhoods.

Implementations

impl GraphSage

pub fn new(config: GraphSageConfig) -> Result<Self>

Create a new GraphSAGE model with the given configuration

pub fn embed(&self, graph: &GraphData) -> Result<GraphSageEmbeddings>

Generate embeddings for all nodes via inductive forward pass.

Performs K-hop neighborhood aggregation where K is the number of layers.

pub fn train_unsupervised( &mut self, graph: &GraphData, ) -> Result<GraphSageTrainingMetrics>

Train the model with unsupervised random-walk loss.

Uses a simple positive/negative sampling strategy where:

• Positive pairs: nodes connected by an edge (BFS neighbors)
• Negative pairs: randomly sampled unconnected nodes

Loss: -log(sigma(pos_score)) - log(1 - sigma(neg_score))

pub fn normalize(v: &[f64]) -> Vec<f64>

L2 normalize a vector

Trait Implementations

impl Clone for GraphSage

fn clone(&self) -> GraphSage

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for GraphSage

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.