yifanfeng97/Hyper-Extract

Hypergraph is more powerful. Transform unstructured text into structured knowledge with LLMs. Graphs, hypergraphs, and spatio-temporal extractions — with one command.

What it solves

Hyper-Extract is a CLI tool designed to transform unstructured documents (PDFs, Word, PowerPoint, Excel, HTML, EPUB) into structured knowledge bases. It eliminates the need to manually read through large volumes of text to find specific facts, allowing users to extract, query, and visualize information as structured data like graphs or lists.

How it works

The tool uses a three-layer architecture consisting of Auto-Types (9 strongly-typed data structures), Methods (extraction and retrieval algorithms like GraphRAG and LightRAG), and Templates (80+ YAML presets for various domains). It leverages LLMs with function calling capabilities to parse documents based on these templates and store them in a queryable index. It also supports incremental updates, allowing users to feed new documents into an existing knowledge base and automatically roll back old facts when sources are updated.

Who it’s for

  • Researchers who need to turn academic papers into interactive concept graphs.
  • Financial Analysts who want to extract entities and metrics from earnings reports.
  • Knowledge Base Managers who need to maintain a synced, traceable knowledge base from evolving documents.
  • Developers who want to integrate structured extraction via a Python API or an MCP server for AI assistants.

Highlights

  • Diverse Knowledge Structures: Supports 9 types, including Hypergraphs, Spatio-Temporal Graphs, and standard Knowledge Graphs.
  • Extensive Template Library: Over 80 zero-code YAML templates across finance, legal, medical, and general domains.
  • Incremental Evolution: Tracks provenance of every fact, enabling source-based updates, rollbacks, and audits.
  • वाइन-compatible Providers: Works with OpenAI, Anthropic, Google Gemini, DeepSeek, and local vLLM deployments.
  • Obsidian Export: Converts extracted graphs into Markdown notes with wikilinks for use in Obsidian vaults.

Related

  • Project
    harshankur/officeParser

    A robust Node.js and browser library that parses diverse office documents into a unified AST for conversion, templating, and RAG-ready data extraction.

  • Project
    Open-Source-Legal/OpenContracts

    An open-source document intelligence platform that transforms document repositories into programmable citation graphs with AI agents, human annotation, and MCP server support.

  • Project
    Egonex-AI/Understand-Anything

    A multi-agent tool that turns codebases and documentation into interactive knowledge graphs, helping developers visualize structure, dependencies, and business logic.

  • Project
    docling-project/docling-parse

    Docling Parse is a fast, configurable PDF‑parsing library (Python API with a C++ core) that extracts text, words, lines, vector shapes and bitmap images together with their coordinates. It supports both sequential and multi‑threaded parsing, lets you choose what to compute or materialize per page, and includes rendering utilities. Ideal for AI workflows that need layout‑aware document data.

  • Project
    firecrawl/anydoc

    A high-performance Rust library that converts various office document formats into clean, consistent Markdown for LLM ingestion and agentic workflows.