openodke

An open implementation of ODKE+ (Apple, arXiv:2509.04696) — independent, and not affiliated with Apple.

The seam between text and any graph store. openodke turns documents — prose, tables, records — into a knowledge graph held to your ontology, and writes it to Neo4j, to JSON Lines, or to a store of your own. It is the only pipeline that asks a second model whether the cited span supports the claim, and records the verdict on every fact.

Every fact carries the document, the character span, the source tier and the grounding verdict that produced it, so why is this edge in my graph? is a query, not an investigation.

Status: v0.1, not yet on PyPI. The data model, ontology I/O, loaders and extraction, grounding and corroboration, the Neo4j sink, odke run and the evaluation harness are on main and tested (milestones M0–M4 and M6 in ROADMAP.md). Much of v0.2 has landed on main too — HTML, PDF and DOCX loaders, RDF, NetworkX and bulk sinks, OWL and Neo4j ontology import. The PyPI release (M7) waits on Trusted Publishing; until then, install from GitHub.

Quickstart

From a clone, with no key, no network and no database:

git clone https://github.com/deepskandpal/odke && cd odke
pip install -e ".[yaml]"
odke run examples/e2e/odke.yaml

That runs a small invented corpus through all thirteen stages on recorded model responses and writes the graph to examples/e2e/out/. The same run from Python:

from openodke import HybridExtractor, LLMExtractor, Ontology, Pipeline, VerdictValidator
from openodke.ground import LLMGrounder
from openodke.llm import RecordedClient, ReplayClient
from openodke.loaders import DirectoryLoader
from openodke.sinks import JsonlSink

ontology = Ontology.from_json("examples/e2e/ontology.json")
docs = list(DirectoryLoader().load("examples/e2e/corpus"))
# Recorded responses, so this runs with no key. Drop `client=` to call the models.
extract = ReplayClient("examples/e2e/recorded/extract.json")
ground = RecordedClient.from_fixture("examples/e2e/recorded/ground.json")
kg = Pipeline(
    ontology,
    HybridExtractor(LLMExtractor(client=extract), documents=docs),
    grounder=LLMGrounder(client=ground),  # a second model checks every cited span
    validator=VerdictValidator(),  # refuses a fact its own span contradicts
    sinks=[JsonlSink("out")],
).run(docs)
print(len(kg.facts), "facts,", kg.stats["refused"], "refused")

examples/e2e/ goes on into Neo4j and ends with the Cypher that shows provenance, a DIFFERENT link between two companies with one name, and a cardinality check.

Does it work?

On the example, odke eval ablation runs one config three ways against 36 labelled facts:

This is a demonstration on hand-authored recorded responses, not a benchmark. The responses and the labels were written for the example, so the table shows what the harness reports, not how well any model does. On that fixture, grounding caught one of four wrong candidates, and normalisation did more for precision than grounding. No number from real model runs exists yet. The paper's 98.8% is neither reproduced nor claimed. What the example shows, and does not.

The harness is the point: run the same command on a slice of your own corpus that you have labelled, and it tells you whether grounding earns its calls there.

Why this exists

Apple published ODKE+ (arXiv:2509.04696), a production system that extracts open-domain facts at scale and ingests them into a knowledge graph at 98.8% precision. Apple released no implementation, and nothing on PyPI, conda, GitLab or Hugging Face implements it. This is an independent implementation of that architecture, generalised from one company's internal graph into an SDK anyone can install.

The three ideas worth taking from that paper:

Other libraries turn text into a graph — iText2KG, neo4j-graphrag's SimpleKGPipeline, LangChain's LLMGraphTransformer. They extract, and most can be handed a schema. None of them has a separate grounding pass, cross-source corroboration, or character-span provenance carried through to the store.

Thirteen stages, each one a Protocol

The pipeline is a superset: thirteen stages, each a typing.Protocol in openodke.stages with a pass-through default (DECISIONS #20). Take the subset your corpus needs, supply your own for any stage by writing one method, and leave the rest out. Nothing domain-specific is in the code; it arrives as an ontology, a config, or a stage of yours.

Install

Not on PyPI yet. Install from GitHub, naming extras the same way:

pip install "openodke @ git+https://github.com/deepskandpal/odke"                # the base install
pip install "openodke[yaml] @ git+https://github.com/deepskandpal/odke"         # + YAML configs
pip install "openodke[neo4j,yaml] @ git+https://github.com/deepskandpal/odke"   # + the Neo4j sink
pip install "openodke[all] @ git+https://github.com/deepskandpal/odke"          # everything

The base install is pydantic and typer, and talks to nothing: the data model, the ontology compiler, the text, Markdown, HTML and record loaders, the pattern extractor, the OpenAI-compatible model client, corroboration, evaluation, and the JSONL and Cypher-file sinks. Each extra is imported on first use, and a missing one is named in the error.

Python 3.11–3.13.

odke run

One file names the inputs, the ontology, the models, which implementation fills each stage, and the sink:

ontology: ontology.json
inputs:
  - path: corpus/register.csv
    loader: {use: csv, tier: curated}
  - corpus/notes
models:
  extract: anthropic/claude-sonnet-5
  ground: anthropic/claude-haiku-4-5-20251001
stages:
  chunker: {use: sentence, max_words: 120}
  extractor: hybrid
  grounder: llm
  normalizer: value
  corroborator: signature
  validator: verdict
  sink: {use: neo4j, uri_env: NEO4J_URI, password_env: NEO4J_PASSWORD}
  constrainer: neo4j
bootstrap: true

odke run config.yaml --dry-run     # load, extract and ground; print what would be written
odke run config.yaml

A stage of your own is package.module:Name with its options alongside. A dry run opens no sink, so it needs no database. Every stage's own counts — verdicts, rejections, conflicts, refusals, cost — go into the graph's stats, and a document's id is its path, so a label or a query can name it. examples/run.yaml comments every key.

Evaluate against your own labels

openodke ships the formats and the arithmetic for scoring every stage, and never a corpus: a number computed against the pipeline's own output measures nothing. Bring your own labelled dataset.

odke eval extract --describe                           # what to label
odke eval extract --labels gold.jsonl --predictions out/facts.jsonl
odke eval ablation --config config.yaml --labels gold.jsonl

There is an evaluator for routing, extraction, grounding, resolution, scoring (Brier, reliability, ECE) and validation, and openodke.eval.CostMeter measures tokens and USD per stage without touching a stage. The fixtures in the test suite exercise the arithmetic and are not a benchmark.

When the platform does a stage itself

Some stores already resolve, prune or constrain after the write. A sink declares that with a PlatformProfile; a stage configured on both sides raises one DoubleStageWarning and still runs, because the two passes are not the same pass (DECISIONS #21). To hand a stage to the platform, pass Delegated(to="neo4j-graphrag:FuzzyMatchResolver") — or {use: delegated, to: ...} in a config — and openodke stamps who did it wherever the data model has room, so the platform's work can be read back and scored with the same evaluator.

Models and providers

Nothing in this library imports a provider SDK or names a vendor. Every model call goes through one LLMClient protocol, and which client serves a model string is decided by a small, inspectable routing rule you can override.

from openodke.llm import ModelRoles, ModelSpec

ModelRoles()  # Claude by default: Sonnet extracts, Haiku grounds
ModelRoles.single("ollama/llama3.1")  # entirely local — no extras, no key, no network
ModelRoles(
    extract=ModelSpec(model="anthropic/claude-sonnet-5"),
    ground=ModelSpec(model="ollama/qwen2.5:3b"),
)

Two models, not one, by default. Extraction wants a capable model; grounding asks thousands of yes/no questions and wants a small one. Anything speaking the OpenAI chat shape — Ollama, vLLM, LM Studio, llama.cpp, OpenRouter, Groq, a gateway — works on the base install; everything else goes through litellm with [llm]; openodke.llm.register("mycorp", factory) routes a provider through your own client. ReplayClient, RecordedClient and ScriptedClient ship in the package, so your own stages are testable offline too.

Ontologies

from openodke import Ontology

ontology = Ontology.from_json("examples/e2e/ontology.json")  # or from_owl, from_neo4j
print(ontology.validate())  # diagnostics with dotted paths, never an exception
print(ontology.snippet("Company").render())  # exactly what the model is prompted with

odke ontology validate schema.json
odke ontology diff old.json new.json --fail-on-breaking
odke ontology snippet schema.json Company

A bad extraction is usually a bad snippet, and snippet shows it before a token is spent. The ontology also marks which qualifiers bear identity (percentile) and which reconcile (start_time), and that decides what counts as one claim.

Honest limits

• The fixtures and the example are not benchmarks. Their model responses are hand-authored. There are no real-model numbers yet.
• No PyPI release yet. M7 waits on Trusted Publishing; install from GitHub.
• Live Neo4j needs 5.7 or later. The constraint bootstrap uses relationship uniqueness constraints; Community edition is enough.
• Structured facts are grounded against their own cell. The cell Leeds cannot say whose head office it is, so a careful grounder answers not_found and the call is spent for little. The default gate keeps not_found for this reason.
• The grounder sees the cited span, never the document. That is what keeps it cheap, and a span that leaves out the subject cannot support the claim.
• Resolution never merges on names. Only a shared identifier re-keys an entity; a name match is a SIMILAR link for someone to act on.
• Inference drafts; it never decides. odke ontology infer proposes a small schema for a person to review and freeze, and nothing infers implicitly (DECISIONS #8).

Documentation

• deepskandpal.github.io/odke — the documentation site: concepts, ontology and inference, loaders and extraction, grounding, resolution and corroboration, sinks, odke run, evaluation
• examples/ — the end-to-end example and the commented run config
• ROADMAP.md — milestones, what each one delivers, and the estimate
• CHANGELOG.md — what landed, milestone by milestone
• DECISIONS.md — the design calls and why they went that way
• CONTRIBUTING.md — ./scripts/verify.sh is the whole check
• NOTICE — the relationship to the ODKE+ paper

Relationship to the paper

This is an independent implementation of a published architecture. It is not affiliated with or endorsed by Apple Inc., uses no Apple code, data or models, and was written from the paper alone. All credit for the architecture belongs to Khorshidi et al. See NOTICE.

It departs from the paper where the paper is specific to its deployment. ODKE+ decides what to refresh by watching Wikipedia edits and retrieves its own evidence; here those two stages are optional protocols, Initiator and Retriever, because an SDK is usually handed its documents.

License

Apache-2.0.