Struct ClusteringService 

pub struct ClusteringService { /* private fields */ }

Service for clustering embeddings.

Supports HDBSCAN and K-means clustering algorithms with automatic prototype computation and cluster quality metrics.

Implementations

impl ClusteringService

pub fn new(config: ClusteringConfig) -> Self

Create a new clustering service with the given configuration.

pub fn default_service() -> Self

Create with default configuration.

pub async fn run_hdbscan( &self, embeddings: &[EmbeddingWithId], ) -> Result<Vec<Cluster>>

Run HDBSCAN clustering on the provided embeddings.

Arguments

• embeddings - Slice of (EmbeddingId, vector) tuples

Returns

A vector of discovered clusters.

pub async fn run_kmeans( &self, embeddings: &[EmbeddingWithId], k: usize, ) -> Result<Vec<Cluster>>

Run K-means clustering on the provided embeddings.

Arguments

• embeddings - Slice of (EmbeddingId, vector) tuples
• k - Number of clusters

Returns

A vector of k clusters.

pub async fn assign_to_nearest( &self, embedding: &[f32], clusters: &[Cluster], ) -> Result<ClusterId>

Assign an embedding to the nearest cluster.

Arguments

• embedding - The embedding vector to assign
• clusters - Available clusters to assign to

Returns

The ID of the nearest cluster.

pub async fn compute_prototypes( &self, cluster: &Cluster, embeddings: &HashMap<EmbeddingId, Vec<f32>>, ) -> Result<Vec<Prototype>>

Compute prototypes (exemplars) for a cluster.

Arguments

• cluster - The cluster to compute prototypes for
• embeddings - All embeddings with their IDs

Returns

A vector of prototypes for the cluster.

pub async fn cluster_with_metrics( &self, embeddings: &[EmbeddingWithId], ) -> Result<ClusteringResult>

Run full clustering pipeline with metrics.