Skip to main content
  • Sync
  • Async
def create_graph(
    name: str,
    filters: Optional[Dict[str, Any]] = None,
    documents: Optional[List[str]] = None,
    prompt_overrides: Optional[Union[GraphPromptOverrides, Dict[str, Any]]] = None,
) -> Graph

​
Parameters

  • name (str): Name of the graph to create
  • filters (Dict[str, Any], optional): Optional metadata filters to determine which documents to include
  • documents (List[str], optional): Optional list of specific document IDs to include
  • prompt_overrides (GraphPromptOverrides | Dict[str, Any], optional): Optional customizations for entity extraction and resolution prompts

​
Returns

Calling create_graph now returns a placeholder Graph immediately.
  • graph (Graph): Graph stub with system_metadata["status"] = "processing". Entities and relationships will be empty until processing completes.
Use db.wait_for_graph_completion("<name>") (sync) or await db.wait_for_graph_completion("<name>") (async) to block until the graph is done, or poll graph.is_processing / graph.is_completed.

​
Examples

  • Sync
  • Async
from morphik import Morphik

db = Morphik()

# Start graph creation – returns immediately with status "processing"
graph = db.create_graph(
    name="research_graph",
    filters={"category": "research"},
)

# Option 1: Block until finished
graph = db.wait_for_graph_completion("research_graph")

# Option 2: Poll periodically
while graph.is_processing:
    time.sleep(10)
    graph = db.get_graph("research_graph")
print("Entities:", len(graph.entities))

# Create a graph from specific documents
graph = db.create_graph(
    name="custom_graph",
    documents=["doc1", "doc2", "doc3"]
)

# With custom entity extraction examples
from morphik.models import EntityExtractionPromptOverride, EntityExtractionExample, GraphPromptOverrides

# Example with only entity extraction examples
graph = db.create_graph(
    name="medical_graph", 
    filters={"category": "medical"},
    prompt_overrides=GraphPromptOverrides(
        entity_extraction=EntityExtractionPromptOverride(
            examples=[
                EntityExtractionExample(label="Insulin", type="MEDICATION"),
                EntityExtractionExample(label="Diabetes", type="CONDITION")
            ]
        )
    )
)

# Example with custom entity extraction prompt template and examples
graph = db.create_graph(
    name="financial_graph",
    documents=["doc1", "doc2"],
    prompt_overrides=GraphPromptOverrides(
        entity_extraction=EntityExtractionPromptOverride(
            prompt_template="Extract financial entities from the following text:\n\n{content}\n\nFocus on these types of entities:\n{examples}\n\nReturn in JSON format.",
            examples=[
                EntityExtractionExample(label="Apple Inc.", type="COMPANY", properties={"sector": "Technology"}),
                EntityExtractionExample(label="Q3 2024", type="TIME_PERIOD"),
                EntityExtractionExample(label="Revenue Growth", type="METRIC")
            ]
        ),
        entity_resolution=EntityResolutionPromptOverride(
            examples=[
                EntityResolutionExample(
                    canonical="Apple Inc.", 
                    variants=["Apple", "AAPL", "Apple Computer"]
                )
            ]
        )
    )
)

print(f"Created graph with {len(graph.entities)} entities and {len(graph.relationships)} relationships")

​
Graph Properties

The returned Graph object has the following properties:
  • id (str): Unique graph identifier
  • name (str): Graph name
  • entities (List[Entity]): List of entities in the graph
  • relationships (List[Relationship]): List of relationships in the graph
  • metadata (Dict[str, Any]): Graph metadata
  • document_ids (List[str]): Source document IDs
  • filters (Dict[str, Any], optional): Document filters used to create the graph
  • created_at (datetime): Creation timestamp
  • updated_at (datetime): Last update timestamp
  • owner (Dict[str, str]): Graph owner information
⌘I