Struct ClusterMap 

pub struct ClusterMap { /* private fields */ }

Pre-computed clustering of feature vectors for fast semantic grouping.

Implementations

impl ClusterMap

pub fn new(dimension: usize) -> Self

Create a new, empty cluster map.

pub fn build(&mut self, nodes: &[(u64, &[f32])], max_iterations: usize)

Run k-means clustering on all feature vectors in the graph. k = min(sqrt(node_count), 256). Skips if node_count < 4.

pub fn nearest_cluster(&self, query: &[f32]) -> Option<usize>

Find the nearest cluster for a query vector.

pub fn get_cluster(&self, cluster_index: usize) -> &[u64]

Get all node IDs in a specific cluster.

pub fn centroid(&self, cluster_index: usize) -> Option<&[f32]>

Get the centroid for a cluster.

pub fn cluster_count(&self) -> usize

Number of clusters.

pub fn assign_node(&mut self, node_id: u64, feature_vec: &[f32])

Assign a new node to the nearest cluster without rebuilding.

pub fn clear(&mut self)

Clear the cluster map.

pub fn is_empty(&self) -> bool

Whether the cluster map is empty.

pub fn dimension(&self) -> usize

Get the dimension.

pub fn centroids(&self) -> &[Vec<f32>]

Get centroids (for serialization).

pub fn assignments(&self) -> &[Vec<u64>]

Get assignments (for serialization).

Trait Implementations

impl Default for ClusterMap

fn default() -> Self

Returns the “default value” for a type.