KnowledgeGraphOptions
Options & Configuration · AiDotNet.RetrievalAugmentedGeneration.Graph
Configuration options for advanced knowledge graph capabilities including embeddings, community detection, link prediction, temporal queries, and KG construction.
For Beginners
After setting up your knowledge graph via ConfigureRetrievalAugmentedGeneration(),
use ConfigureKnowledgeGraph() to enable advanced features:
- Train embeddings to enable link prediction and entity similarity
- Detect communities for global search capabilities
- Enable temporal queries for time-aware reasoning
- Construct a KG automatically from text input
How It Works
These options are separate from ConfigureRetrievalAugmentedGeneration(), which handles
low-level plumbing (IGraphStore, KnowledgeGraph, HybridGraphRetriever). This class configures
higher-level features built on top of the existing infrastructure.
Constructors
| Constructor | Summary |
|---|---|
KnowledgeGraphOptions() |
Properties
| Property | Summary |
|---|---|
GraphRAGMode? GraphRAGMode { get; set; } |
GraphRAG retrieval mode. |
GraphRAGOptions GraphRAGOptions { get; set; } |
Options for GraphRAG retrieval. |
KGConstructionOptions ConstructionOptions { get; set; } |
Options for KG construction from text. |
KGEmbeddingOptions EmbeddingOptions { get; set; } |
Options for embedding model training. |
KGEmbeddingType? EmbeddingType { get; set; } |
Type of embedding model to use. |
List<string> ConstructionTexts { get; set; } |
Text documents to construct the knowledge graph from. |
bool? EnableLinkPrediction { get; set; } |
Whether to enable link prediction. |
bool? TrainEmbeddings { get; set; } |
Whether to train knowledge graph embeddings. |
double? LinkPredictionTestFraction { get; set; } |
Fraction of edges to hold out for link prediction evaluation. |
int? LinkPredictionMaxTestEdges { get; set; } |
Maximum number of test edges for link prediction evaluation. |