ontograph 0.4.0

LLM-powered knowledge graph engine with ontological entity resolution, orbit-based proximity scoring, and hybrid semantic-graph retrieval.

ontograph

LLM-powered knowledge graph engine
Ontological entity resolution · Orbit-based proximity scoring · Hybrid semantic-graph retrieval

---

Install

pip install ontograph

Requires OPENAI_API_KEY environment variable (or set it in a .env file — loaded automatically).

Quick Start

from ontograph import OntoDB, Schema

db = OntoDB("my_knowledge.db")

# Define an ontology schema (constrains what the LLM extracts)
db.register_schema(Schema(
    name="workplace",
    entity_types=["person", "project", "team", "meeting", "topic"],
    relationship_types=[
        {"name": "works_on", "directed": True},
        {"name": "manages", "directed": True},
        {"name": "colleague", "directed": False},
        {"name": "discussed", "directed": True},
    ],
))

# Ingest unstructured text — entities and relationships extracted automatically
db.ingest(
    "Meeting with Nara and Marco about Project Neptune. "
    "Nara is leading the backend rewrite. Marco raised concerns about the launch deadline.",
    source_type="transcript",
    schema_name="workplace",
)

# Search the graph (semantic + keyword + graph traversal)
results = db.search("What is Nara working on?")

# Ask a question — LLM synthesizes answer from graph context
answer = db.ask("Who is concerned about the launch deadline?")
print(answer)

Core Concepts

Entities, Relationships, Attributes

Everything decomposes into three primitives:

Primitive	Description
Entities	Nodes — people, projects, topics, anything nameable
Relationships	Edges — directed (A → B) or bidirectional (A ↔ B)
Attributes	Key-value metadata on entities and relationships

Entity Resolution

When ingesting text, names are fuzzy-matched against existing entities using a four-signal composite score:

Signal	Method	Catches
Phonetic	Metaphone	Pronunciation-similar names
Spelling	Jaro-Winkler	Typos and minor variations
Semantic	Embedding cosine	Conceptual matches
Orbit	Interaction frequency	Context-aware disambiguation

Orbit

Your "orbit" is a proximity model. Entities you interact with frequently score higher — when "Sal" appears in a transcript, the system weights your manager "Sam" (who you interact with daily) over random "Sal"s elsewhere.

sam = db.get_entity("Sam")
db.add_alias(sam.id, "Sal", alias_type="transcript_error")

File References

Entities can reference external files — receipts, photos, PDFs, contracts — that live on disk. These references are surfaced during search and Q&A so the LLM knows where to find supporting material.

# Attach files to an entity
db.attach_files("kitchen renovation", [
    "/Users/me/photos/kitchen_before.jpg",
    "/Users/me/photos/kitchen_after.jpg",
])

# When you ask a question, the LLM sees the file references
answer = db.ask("What photos do we have of the kitchen?")

# Or via CLI
ontograph attach "kitchen renovation" /Users/me/photos/kitchen_before.jpg

Schemas

Ontology schemas define valid entity and relationship types for a domain. They constrain what the LLM can extract during ingestion.

Self-Improving Feedback Loop

Every entity resolution is logged. Mark resolutions as correct/incorrect to track accuracy over time:

db.mark_resolution("log_id", correct=True)
print(db.stats()["resolution_accuracy"])

Python API

Method	Description
OntoDB(db_path, api_key, observer_id)	Main entry point
.ingest(text, source_type, schema_name)	Ingest unstructured text
.search(query, limit, graph_depth)	Hybrid search
.ask(question)	LLM-synthesized answer from graph context
.add_entity(name, type, attributes, file_refs)	Manual entity creation
.add_relationship(source, target, type)	Manual relationship creation
.attach_files(entity, file_paths)	Attach file references
.detach_files(entity, file_paths)	Remove file references
.resolve(name, entity_type, observer)	Entity resolution
.orbit(observer, limit)	Proximity-ranked entities
.stats()	Graph statistics and resolution accuracy

CLI

ontograph --help                    # Full command list with examples
ontograph ingest --text "..."       # Ingest text (also accepts --file or stdin)
ontograph search "query"            # Hybrid search
ontograph ask "question"            # LLM-synthesized answer
ontograph entities [--type person]  # List entities
ontograph entity "name"             # Get single entity
ontograph relationships "name"      # Get relationships
ontograph neighbors "name"          # Graph traversal
ontograph resolve "name"            # Entity resolution
ontograph attach "entity" file.pdf  # Attach file references
ontograph detach "entity" file.pdf  # Remove file references
ontograph schema register --file s.json  # Register ontology schema
ontograph dashboard                 # Interactive graph visualization

Dashboard

ontograph dashboard                 # Opens browser at http://127.0.0.1:8484
ontograph dashboard --port 9000    # Custom port
ontograph --db project.db dashboard

Interactive force-directed graph visualization powered by D3.js. Starts with the most-connected entity as hub, showing its top 50 relationships. Click to select, double-click to expand connections on demand, drag to reposition, scroll to zoom.

License

MIT