Get concepts related to a given concept

Get relevance score for a term # Arguments `term` - The term to check # Returns Relevance score between 0.0 and 1.0

Get the root nodes of the tree

Get a cloned sender for submitting operations

Get current statistics

Get statistics about enrichment

Get total requests

Guess entity type based on simple heuristics

Handle executing a query in Explain mode (returns confidence + sources)

Get cache hit rate as a percentage (0.0 to 1.0)

Import CSV/TSV with custom delimiter

Increments the counter for failed document processing operations

Increments the counter for successfully processed documents

Index a document by extracting concepts and building the graph This processes the document, extracts concepts, and adds them to the graph builder. Ca

Initialize quality checks

Initialize safety patterns

Get the underlying LLM client

Determines invalidation strategies for a set of changes

Check if this is a causal relationship type

Check if a word is a common word that shouldn't be extracted as an entity

Check if the store is empty

Check if system is initialized

Check if a phrase is a valid concept

Check if visible

Calculate Jaccard similarity between two texts

Calculate Levenshtein-based similarity

Compute Levenshtein edit distance-based similarity

List all workspaces

Load workspace metadata

Load workspace metadata

Make HTTP GET request

Make HTTP POST request

Mark an edge as dirty

Mark a node as dirty

Set the maximum number of queries to warm

Get memory usage in bytes

Concatenate results (first, then second)

Get workspace metadata file path

Get model information with "Cached" prefix

Get the name of the operation being timed

Normalize weights to sum to 1.0

Normalize weights to [0, 1] range using min-max normalization

Normalize whitespace by collapsing multiple spaces and trimming

Optimize a query operation

Parse a slash command from input string

Parse dependencies (simplified)

Compute phonetic similarity using simplified Soundex

Plan entity search strategy

Pre-compute global PageRank scores for faster query processing

Print query results

Print pipeline statistics

Process a single document

Process a single file into document metadata

Get propagation statistics

Query the system for relevant information

Query with document filter

Records the processing duration of a document Keeps only the most recent 1000 measurements to prevent unbounded memory growth. # Parameters - `durat

Record relevance feedback # Arguments `term` - The search term `score` - Relevance score (0.0 to 1.0)

Reload configuration from source

Set retrieval strategy

Run the full GraphRAG CLI. Called by both the `graphrag-cli` binary and the `graphrag` meta-crate binary.

Run benchmark with given configuration

Sanitize string by removing control characters

Save workspace metadata

Save workspace metadata

Save knowledge graph to JSON file with optimized format for entities and relationships

Score a sentence based on multiple factors Scoring criteria: 1. Position score (beginning and end sentences are important) 2. Length score (prefer me

Search for similar documents # Arguments `query_embedding` - Query embedding vector `limit` - Maximum number of results `filter` - Optional filter on

Add attribute to span

Set in L1 cache

Check if propagation should be triggered

Get number of relationships in cluster

Simple Soundex implementation

Get Soundex code for a character

Split an oversized document into sections that fit within the context window Sections are formed by grouping paragraphs (double-newline boundaries) u

Split text into sentences using simple sentence tokenization

Starts tracking a new operation and returns its ID

Get graph statistics