Struct SemanticRouter 

pub struct SemanticRouter { /* private fields */ }

Semantic router combining CID-based and vector-based search

Provides intelligent content discovery through vector similarity search over content embeddings.

Implementations

impl SemanticRouter

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

Create a new semantic router with the given configuration

pub fn with_defaults() -> Result<Self>

Create a new router with default configuration

pub fn add(&self, cid: &Cid, embedding: &[f32]) -> Result<()>

Add content with its embedding to the router

Arguments

• cid - Content identifier
• embedding - Vector embedding of the content

pub fn add_batch(&self, items: &[(Cid, Vec<f32>)]) -> Result<()>

Add multiple content items in batch

More efficient than adding one by one

Arguments

• items - Vector of (CID, embedding) pairs

pub fn remove(&self, cid: &Cid) -> Result<()>

Remove content from the router

pub fn contains(&self, cid: &Cid) -> bool

Check if content exists in the router

pub async fn query( &self, query_embedding: &[f32], k: usize, ) -> Result<Vec<SearchResult>>

Query for content by semantic similarity

Arguments

• query_embedding - Query vector
• k - Number of results to return

pub async fn query_auto( &self, query_embedding: &[f32], k: usize, ) -> Result<Vec<SearchResult>>

Query with auto-tuned ef_search parameter

Automatically determines the optimal ef_search based on k and index size

Arguments

• query_embedding - Query vector
• k - Number of results to return

pub async fn query_with_ef( &self, query_embedding: &[f32], k: usize, ef_search: usize, ) -> Result<Vec<SearchResult>>

Query with custom ef_search parameter

Arguments

• query_embedding - Query vector
• k - Number of results to return
• ef_search - Search parameter (higher = more accurate but slower)

pub async fn query_with_filter( &self, query_embedding: &[f32], k: usize, filter: QueryFilter, ) -> Result<Vec<SearchResult>>

Query with filtering options

Arguments

• query_embedding - Query vector
• k - Number of results to return
• filter - Query filter options

pub fn clear_cache(&self)

Clear the query result cache

pub fn cache_stats(&self) -> CacheStats

Get cache statistics

pub fn stats(&self) -> RouterStats

Get statistics about the router

pub fn optimization_recommendations(&self) -> OptimizationRecommendations

Get optimization recommendations

Returns recommended HNSW parameters for the current index size

pub async fn save_index<P: AsRef<Path>>(&self, path: P) -> Result<()>

Save the semantic index to a file

Serializes the entire HNSW index including all vectors and CID mappings to a file for later loading.

Arguments

• path - Path to save the index file

pub async fn load_index<P: AsRef<Path>>(&self, path: P) -> Result<()>

Load a semantic index from a file

Loads a previously saved HNSW index from disk, replacing the current index.

Arguments

• path - Path to the saved index file

pub fn clear(&self) -> Result<()>

Clear all content from the router

pub async fn query_with_aggregations( &self, query_embedding: &[f32], k: usize, filter: QueryFilter, ) -> Result<(Vec<SearchResult>, SearchAggregations)>

Query with aggregations

Returns both results and aggregated statistics

Arguments

• query_embedding - Query vector
• k - Number of results to return
• filter - Query filter options

pub async fn query_batch( &self, query_embeddings: &[Vec<f32>], k: usize, ) -> Result<Vec<Vec<SearchResult>>>

Batch query for multiple embeddings at once

More efficient than querying one by one due to parallelization and amortized overhead.

Arguments

• query_embeddings - Multiple query vectors
• k - Number of results to return per query

Returns

Vector of search results, one for each query in the same order

pub async fn query_batch_with_filter( &self, query_embeddings: &[Vec<f32>], k: usize, filter: QueryFilter, ) -> Result<Vec<Vec<SearchResult>>>

Batch query with filtering options

Processes multiple queries in parallel with filtering applied to each.

Arguments

• query_embeddings - Multiple query vectors
• k - Number of results to return per query
• filter - Query filter options (applied to all queries)

Returns

Vector of search results, one for each query in the same order

pub async fn query_batch_with_ef( &self, query_embeddings: &[Vec<f32>], k: usize, ef_search: usize, ) -> Result<Vec<Vec<SearchResult>>>

Batch query with custom ef_search parameter

Processes multiple queries in parallel with custom search parameter.

Arguments

• query_embeddings - Multiple query vectors
• k - Number of results to return per query
• ef_search - Search parameter (higher = more accurate but slower)

Returns

Vector of search results, one for each query in the same order

pub fn batch_stats(&self, batch_results: &[Vec<SearchResult>]) -> BatchStats

Get batch query statistics

Returns aggregated statistics across a batch of queries

Trait Implementations

impl Default for SemanticRouter

fn default() -> Self

Returns the “default value” for a type.