Knowledge graph extractor
Extract entities and relationships from text, GraphRAG-style, as a Mermaid graph, JSON triples or tables with evidence.
Your key stays in your browser.It goes straight to the provider, never to our servers, and we neither log nor keep it. Use a dedicated key with a spending cap, and delete it in the provider's dashboard when you are done.
Input
Result
The result will appear here.Microsoft's GraphRAG showed that a knowledge graph built by a language model — entities, their types, and typed relationships with a strength score — lets retrieval answer questions that no single chunk contains. The first step of that pipeline is extraction, and it is useful on its own: from an article, a contract, a project history or a set of meeting notes, pull out who is who, what belongs to what, and who did what to whom, with the sentence that proves it. This tool runs that step on one document and renders the result as a Mermaid graph you can paste into a README, as JSON triples for a graph database, or as tables for review.
How it works
- You choose the entity types — people, organisations, places, products, events, concepts, or your own list — and the model extracts only those, resolving pronouns and abbreviations to one canonical name.
- Every relationship carries a strength from 1 to 10 and the words from the text that support it, so a wrong edge can be traced and deleted.
- Nothing outside the text is added: an entity's description collects what the document says and never what the model knows about it.
Where your data goes
With your own key, what you type and the key go from your browser straight to the provider you choose; neither passes through Hysen Labs. In hosted mode, your text goes through our server to our provider (DeepSeek) and is billed in credits; we keep the token counts and cost of each run for billing, never the text itself or the answer. How the provider handles the text is governed by its own privacy policy.
This tool handles keys and credentials, so nothing about a run is saved, not even to your own history.
About your API key
We do not collect, store or leak your key: it lives only in this page's memory (unless you tick “remember in this tab”) and is gone when you close it. Still, treat any key you have pasted into a web page with care — create a dedicated key with a spending cap for use here, and delete or rotate it in your provider's dashboard when you are done.
What it costs
This tool is free, with no sign-in and no points.
Common questions
- How much text can I paste?
- As much as your model's context allows — tens of thousands of words on current models — but extraction quality drops on very long inputs because the model loses track of earlier entities. GraphRAG itself chunks documents at about 1,200 tokens and merges the graphs afterwards; for a long document, extract per section and merge the JSON.
- Can I load the JSON into Neo4j or a vector database?
- The JSON has an entities array and a relationships array with source, target, relationship, strength and evidence, which maps one-to-one onto nodes and edges. For Neo4j, UNWIND the entities into MERGE statements and the relationships into MATCH … MERGE; for a property graph in Python, networkx reads it in a few lines.
- Why are some relationships missing?
- Only relationships the text states or clearly implies are extracted, by design — inferring 'competes with' from two companies being in the same market is the model's knowledge, not the document's. If the text says it and the tool missed it, lower the bar by adding a more specific entity type list.
The open-source behind it
This tool's prompt is adapted from microsoft/graphrag (MIT) and runs on the model you choose. To use the same capability from the command line or inside your own program, start with that project.
microsoft/graphragAlso known as
- knowledge graph extraction from text
- entity relationship extraction llm
- graphrag entity extraction
- text to knowledge graph
- mermaid graph from text
- triples extraction