trustgraph/trustgraph_configurator/templates/1.6/flows/flow-classes.jsonnet
elpresidank 74cc8a4685 Squashed 'ai-context/trustgraph-templates/' content from commit 42a5fd1b
git-subtree-dir: ai-context/trustgraph-templates
git-subtree-split: 42a5fd1b678f32be378062e30451e2052ccb95dd
2026-04-05 21:09:49 -05:00

105 lines
No EOL
4.1 KiB
Jsonnet

// TrustGraph Flow Classes Configuration
// Defines different flow combinations for various use cases
// Each flow class combines multiple functional modules to create complete processing pipelines
//
// Available modules:
// - graphrag: Graph-based RAG with knowledge graphs
// - documentrag: Document-based RAG with chunk embeddings
// - structured: Structured data processing and NLP queries
// - agent: AI agent orchestration and tool integration
// - load: Document loading and preprocessing
// - kg-base: Basic knowledge extraction from text
// - agent-extract: Agent-based knowledge extraction
// - kgcore: Knowledge graph core storage
// Import all the modular flow components
local graphrag_part = import "graphrag.jsonnet";
local kg_base_part = import "kg-base.jsonnet";
local onto_base_part = import "onto-base.jsonnet";
local agent_extract_part = import "agent-extract.jsonnet";
local structured_part = import "structured.jsonnet";
local documentrag_part = import "documentrag.jsonnet";
local agent_part = import "agent.jsonnet";
local load_part = import "load.jsonnet";
local kgcore_part = import "kgcore.jsonnet";
{
// Complete TrustGraph system with all capabilities
// Includes GraphRAG, DocumentRAG, structured data processing, and knowledge cores
"everything": {
description: "GraphRAG, DocumentRAG, structured data + knowledge cores",
tags: [
"document-rag", "graph-rag", "knowledge-extraction",
"structured-data", "kgcore"
],
} +
graphrag_part + documentrag_part + agent_part + load_part +
kg_base_part + structured_part,
// Dual RAG system without knowledge core creation
// Combines both document and graph-based retrieval
"document-rag+graph-rag": {
description: "Supports GraphRAG and document RAG, no core creation",
tags: ["document-rag", "graph-rag", "knowledge-extraction"],
} +
graphrag_part + documentrag_part + agent_part + load_part + kg_base_part,
// Graph-based RAG only
// Uses knowledge graphs for context-aware question answering
"graph-rag": {
description: "GraphRAG only",
tags: ["graph-rag", "knowledge-extraction"],
} +
graphrag_part + agent_part + load_part + kg_base_part,
// Graph-based RAG only
// Uses knowledge graphs for context-aware question answering
"onto-rag": {
description: "Ontology RAG only",
tags: ["graph-rag", "knowledge-extraction"],
} +
graphrag_part + agent_part + load_part + onto_base_part,
// Document-based RAG only
// Uses document embeddings for semantic search and answers
"document-rag": {
description: "DocumentRAG only",
tags: ["document-rag"],
} +
documentrag_part + load_part,
// Full RAG system with knowledge core creation
// Includes both RAG types plus persistent knowledge storage
"document-rag+graph-rag+kgcore": {
description: "GraphRAG + DocumentRAG + knowledge core creation",
tags: ["document-rag", "graph-rag", "knowledge-extraction"],
} +
graphrag_part + documentrag_part + agent_part + load_part +
kgcore_part + kg_base_part,
// GraphRAG with advanced agent-based extraction
// Uses AI agents for sophisticated knowledge extraction
"graph-rag+agent-extract": {
description: "GraphRAG + agent extract",
tags: ["graph-rag", "knowledge-extraction", "agent-extract"],
} +
graphrag_part + agent_part + load_part + agent_extract_part,
// GraphRAG with structured data processing
// Combines knowledge graphs with structured data queries
"graph-rag+structured-data": {
description: "GraphRAG + structured data",
tags: ["graph-rag", "knowledge-extraction", "structured-data"],
} +
graphrag_part + agent_part + load_part + structured_part,
// Structured data processing only
// Handles structured data extraction and NLP queries
"structured-data": {
description: "Structured data only",
tags: ["knowledge-extraction", "structured-data"],
} +
agent_part + load_part + structured_part,
}