Build a complete (non-streaming) response Generates a full response without streaming, using the same synthesis logic as streaming responses but deli

Build hierarchical tree for a document asynchronously

Build prompt for keyword extraction

Build corpus knowledge graph from entity clusters

Build hierarchy from sections

Build hierarchy from sections

Build hierarchy from sections

Build a graph where relationships are nodes

Build sections from headings

Build sections from headings

Build sections from headings

Build semantic similarity connections between entities

Build and validate the warming configuration

Get cache utilization (0.0 to 1.0)

Calculate confidence score for the generated answer

Calculate attractive forces along edges

Calculate average node degree in snapshot

Calculate BLEU score (BiLingual Evaluation Understudy) BLEU score measures n-gram overlap between generated and reference text, with a brevity penalt

Calculate BM25 term score

Calculate centrality scores for all nodes # Returns Map of node ID to centrality scores

Calculate coherence score

Calculate completeness score

Calculate completeness score

Calculate relevance score for a concept based on graph connectivity

Calculate confidence score for a relationship

Calculate confidence score

Calculate content hash for deduplication

Calculate evidence score for a potential relationship Analyzes a text chunk to determine how strongly it suggests a relationship between two entities

Calculate exact match score Score = (matched_tokens / total_query_tokens)

Calculate final confidence score Combines synthesis confidence with intent classification confidence using weighted averaging when confidence weighti

Calculate fuzzy match score using edit distance Uses Levenshtein distance for typo tolerance

Calculate IDF (Inverse Document Frequency) for a term Uses Lucene-style IDF to avoid negative values for common terms

Calculate importance scores for entities

Calculate score for a specific intent

Calculate latency change from enabled features

Calculate linking accuracy (placeholder implementation)

Calculate modularity for a set of nodes

Calculate modularity delta for moving node between communities

Calculate name similarity

Calculate precision for n-grams with clipping

Calculate numeric content score

Calculate overall confidence from matches

Calculate overall confidence from bindings

Calculate PageRank scores for concepts Uses the petgraph PageRank implementation with configurable damping.

Calculate proper noun score

Calculate proximity score between entities in text Determines how close two entities are mentioned in the text. Closer proximity suggests a stronger

Calculate quality boost from enabled features

Calculate quality metrics

Calculate readability score

Calculate relevance score

Calculate relevance score for a node given query keywords

Calculate repulsive forces between all nodes

Calculate ROUGE-L score (Recall-Oriented Understudy for Gisting Evaluation - Longest Common Subsequence) ROUGE-L measures the similarity between cand

High-performance dense matrix computation for small graphs

Parallel sparse computation for large graphs

Calculate semantic similarity using word overlap and semantic relatedness This is a lightweight proxy for true embedding-based semantic similarity. F

Calculate cosine similarity differences between consecutive sentences

Calculate source credibility score

Calculate statistical significance

Calculate matching statistics

Calculate statistics for ranked concepts

Calculate strategy weights based on query analysis

Calculate structure boost for chunk

Calculate current throughput

Calculate the time span of the chain

Calculate token reduction from enabled features

Calculate word frequency score

Execute multiple function calls in sequence

Cancel the background task

Helper to center a rect

Pre-flight check: verify required Ollama models are available. Queries Ollama's `/api/tags` (3-second timeout) and checks that every model the curren

Check for quality alerts

Check temporal consistency of the chain Returns true if all steps are temporally ordered (t1 < t2 < t3...)

Sentence-based chunking

Semantic similarity-based chunking Uses lexical cohesion (word overlap) to measure similarity between sentences. Creates chunk boundaries where simil

Topic-based chunking (simplified TextTiling algorithm)

Fixed-size chunking (baseline)

Hybrid chunking strategy

Split text into chunks with overlap using hierarchical boundary preservation

Chunk text hierarchically and enrich with semantic metadata

Chunk text using any strategy that implements ChunkingStrategy trait This method provides a flexible way to use different chunking approaches while m

Add context chunks

Get all chunks (mutable)