★rXivSI
Get started
    • Home
GraFlo: Universal Data-to-Graph Transformer Framework Logo

GraFlo

Python 3.11PyPI versionPyPI DownloadsLicense: BSLpre-commitDOI

GraFlo: The Universal Data-to-Graph Transformer

Transform tabular (CSV, SQL) and hierarchical (JSON, XML) data into Property Graphs and Knowledge Graphs for seamless ingestion into major graph databases (Neo4j, ArangoDB, TigerGraph). Ideal for building knowledge graphs from structured data sources and integrating with LLM workflows.

🛠️ Stop the Custom Code. Start Graphing.

Instead of building brittle, custom scripts for every dataset and graph database:

Data → Custom Function → Specific Graph

Embrace a universal standard:

Any Data → GraFlo (Declarative Schema) → Any Graph

Stop writing idiosyncratic functions. Define your graph structure once with GraFlo's declarative Schema, and turn Any Data into Any Graph.

Core Concepts

  • Property Graphs: Nodes (vertices) and relationships (edges) with properties and unique identifiers. The standard graph model used by modern graph databases.
  • Declarative Schema: Define how your data is transformed into a graph using a declarative configuration. Specify vertex/edge definitions, resource mapping, and transforms without writing custom code.
  • Resources: Your data sources—table-like (CSV files, SQL databases) or hierarchical (JSON files, XML documents). GraFlo handles parsing and transformation automatically.

Why GraFlo?

  • Universal Standard: One declarative Schema works across all data sources and graph databases
  • Knowledge Graph Construction: Build structured knowledge graphs from structured data sources (CSV, SQL, JSON, XML)
  • LLM Integration: Transform structured data into knowledge graphs ready for LLM-powered applications and RAG systems
  • Multi-Tool Power: Handles parsing CSV/SQL/JSON/XML and defining Vertices, Edges, and Properties
  • Database-Agnostic: Seamless ingestion into Neo4j, ArangoDB, and TigerGraph
  • Declarative Schema: Define your graph structure once, reuse everywhere
  • High Performance: Parallel processing for efficient data transformation
  • Advanced Features: Compound indexes, blank vertices, filtering, and server-side filtering support

Installation

pip install graflo

Requires Python 3.11 or higher. Available on PyPI.

Resources

Complete Knowledge Graph Pipeline

GraFlo is part of GrowGraph's open source knowledge graph ecosystem. For a complete solution:

  • Structured Data? Use GraFlo to transform CSV, SQL, JSON, XML into knowledge graphs
  • Unstructured Documents? Use OntoCast - Agentic LLM-powered framework for building knowledge graphs from unstructured documents

Together, GraFlo and OntoCast provide a complete solution for building knowledge graphs from both structured and unstructured data, perfect for LLM-enhanced workflows and agentic optimization.