hkgb

Bridge CSV or database domain graphs with LLM-extracted entities from PDFs and text in Neo4j.

7|3|Updated Jan 15, 2026

Search Tags:#pipelines #knowledge-graph #neo4j #document-ingestion #graphrag #hybrid-graph #llm-extraction

Ready to run

npx skills add https://github.com/Zpankz/mcp-skillset --skill hkgb

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: hkgb
Source: https://github.com/Zpankz/mcp-skillset/tree/main/hkgb
Command: npx skills add https://github.com/Zpankz/mcp-skillset --skill hkgb

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Bridge structured domain data with lexical graphs extracted from unstructured documents to enable cohesive knowledge graph construction across heterogeneous sources and systems.

Core Features & Use Cases

Bridges domain graphs (CSV/database) with lexically extracted entities from PDFs/text via LLMs
Provides a repeatable pipeline for GraphRAG-style ingestion, metadata enrichment, and cross-graph joins using a common Document.path key
Use Case: Integrate customer data with document-derived references to build a unified knowledge graph in Neo4j

Quick Start

Configure your domain CSV data and PDFs, then run the SimpleKGPipeline to build and connect domain and lexical graphs.

Frequently Asked Questions about hkgb

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build a GraphRAG knowledge graph from both domain CSV data and unstructured PDFs?▼

To build a GraphRAG knowledge graph from mixed sources, use a schema-guided pipeline that extracts LLM-derived entities from PDFs and bridges them with domain CSV data using a common Document.path key for coherent cross-graph joins.

What's the best way to integrate LLM-extracted entities from documents with structured domain data in Neo4j?▼

The best way to integrate LLM-extracted entities with structured domain data in Neo4j is applying a Cypher-based join workflow that bridges lexical graphs from text with domain graphs via a shared Document.path metadata key.

How does the document-path bridging mechanism connect lexical graphs and domain graphs?▼

The document-path bridging mechanism connects lexical graphs and domain graphs by using Document.path as a common join key, enabling metadata enrichment and coherent cross-graph relationships across heterogeneous data sources in Neo4j.

Can I use SimpleKGPipeline to ingest PDFs and connect them to an existing database domain graph?▼

Yes, you can configure the SimpleKGPipeline to ingest PDFs and extract lexical entities, then connect the resulting lexical graph to your existing database domain graph using a Cypher-based join workflow in Neo4j.

Do I need embeddings to construct a knowledge graph from unstructured documents and structured data?▼

Yes, an embedding-backed KG builder is required to construct a unified knowledge graph, as it supports the schema-guided extraction pipeline that bridges domain data with lexical graphs extracted from unstructured documents.

When should I use a hybrid-graph approach for knowledge graph construction instead of a single-source pipeline?▼

Use a hybrid-graph approach for knowledge graph construction when integrating heterogeneous sources like database domain graphs with LLM-derived entities from unstructured PDFs, requiring metadata enrichment and cross-graph joins for GraphRAG enablement.