neo4j-document-import-skill

To ingest unstructured PDF documents into a Neo4j knowledge graph, use SimpleKGPipeline to chunk content into Document and Chunk structures, then extract typed entities and relationships via LLM. Configure a defined GraphSchema to drive schema-driven extraction and store embeddings for graph workloads.

Loading semi-structured JSON data into Neo4j for GraphRAG workflows uses `apoc.load.json` for flexible ingestion. After loading, SimpleKGPipeline extracts entities and relationships from the content using a defined schema, creating connected graph structures suitable for retrieval-augmented generation.

LLM entity extraction for building a knowledge graph from text works by processing chunked Document and Chunk structures through SimpleKGPipeline. The pipeline uses a configured GraphSchema to identify and map typed entities and edges, applying entity resolution to merge duplicates and ensure reliable graph ingestion.

Yes, you can use LangChain document loaders to import text into Neo4j. The ingestion workflow supports LangChain and LlamaIndex document-loader integrations, allowing you to pass unstructured text directly to SimpleKGPipeline for chunking, LLM-driven extraction, and graph node creation.

You do not strictly need a no-code tool to extract entities from Markdown files, as SimpleKGPipeline supports programmatic ingestion. However, Neo4j LLM Graph Builder is available as a no-code alternative for users who want to visually ingest Markdown and text documents into a knowledge graph without writing code.

Limitations of SimpleKGPipeline for document ingestion include its focus on unstructured and semi-structured text rather than structured CSV imports. It handles extraction and entity resolution but avoids retrieval, vector indexing, and structured CSV import responsibilities, which are managed by other specialized skills.