Typical Methods

Direct extraction methods that process text without retrieval.

---

iText2KG

hyperextract.methods.typical.iText2KG

Bases: AutoGraph[NodeSchema, EdgeSchema]

iText2KG: A specialized AutoGraph for extracting high-quality triple-based KGs.

Features: - Fixed Schema (NodeSchema, EdgeSchema) optimized for triple extraction - Customized prompts from original iText2KG* implementation - Automatic deduplication and consistency checking via AutoGraph's OMem - Two-stage extraction: entities first, then relationships

Example

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) embedder = OpenAIEmbeddings() kg = iText2KG(llm_client=llm, embedder=embedder) kg.feed_text(text) print(f"Extracted {len(kg.nodes)} entities and {len(kg.edges)} relationships.")

Functions

__init__(llm_client: BaseChatModel, embedder: Embeddings, chunk_size: int = 2048, chunk_overlap: int = 256, max_workers: int = 10, verbose: bool = False)

Initialize iText2KG.

Parameters:

Name	Type	Description	Default
llm_client	BaseChatModel	Language model for extraction	required
embedder	Embeddings	Embedding model for vector indexing	required
chunk_size	int	Characters per chunk	2048
chunk_overlap	int	Overlapping characters between chunks	256
max_workers	int	Max concurrent extraction workers	10
verbose	bool	Display detailed execution logs and progress information	False

---

iText2KG_Star

hyperextract.methods.typical.iText2KG_Star

Bases: AutoGraph[NodeSchema, EdgeSchema]

iText2KG_Star: A specialized AutoGraph for extracting high-quality triple-based KGs.

Features: - One-stage extraction: Extracts edges directly, then derives nodes automatically - Customized prompts from original iText2KG_Star implementation - Semantic Deduplication: Includes match_nodes_and_update_edges using SemHash with embeddings - Automatic date tracking: Observation date set to extraction time - Nested node schema: Maintains rich semantic information (name + label)

Example

from langchain_openai import ChatOpenAI, OpenAIEmbeddings llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) embedder = OpenAIEmbeddings() kg = iText2KG_Star(llm_client=llm, embedder=embedder)

1. Extract relationships and derive nodes

kg.feed_text("Elon Musk is the CEO of SpaceX. Musk leads Tesla.") print(f"Before dedup - Nodes: {len(kg.nodes)}, Edges: {len(kg.edges)}")

2. Deduplicate: Merges 'Elon Musk' and 'Musk'

kg.match_nodes_and_update_edges(threshold=0.85) print(f"After dedup - Nodes: {len(kg.nodes)}, Edges: {len(kg.edges)}")

Attributes

observation_date = observation_date instance-attribute

Functions

__init__(llm_client: BaseChatModel, embedder: Embeddings, observation_date: str | None = None, chunk_size: int = 2048, chunk_overlap: int = 256, max_workers: int = 10, verbose: bool = False)

Initialize iText2KG_Star.

Parameters:

Name	Type	Description	Default
llm_client	BaseChatModel	Language model for extraction	required
embedder	Embeddings	Embedding model for vector indexing	required
observation_date	str | None	Date when the extraction was performed, like '1997-10-10' or '1997-10-10 23:59:59'. If None, uses current date and time.	None
chunk_size	int	Characters per chunk	2048
chunk_overlap	int	Overlapping characters between chunks	256
max_workers	int	Max concurrent extraction workers	10
verbose	bool	Display detailed execution logs and progress information	False

_extract_data(text: str) -> AutoGraphSchema

Extract edges directly, then derive nodes (One-stage extraction).

Process: 1. Split text into chunks. 2. Batch extract edges directly from chunks (Edges contain node info). 3. Post-process: Set observation date to current date and time. 4. Derive nodes from edges. 5. Merge all partial graphs into one global graph.

Parameters:

Name	Type	Description	Default
text	str	Input text.	required

Returns:

Type	Description
AutoGraphSchema	Extracted and validated graph.

match_nodes_and_update_edges(threshold: float = 0.8) -> iText2KG_Star

Match nodes in the graph and update edges accordingly using SemHash with embeddings.

This method identifies and merges similar nodes based on semantic similarity using embeddings from the instance's embedder. It updates edges to reflect any changes in node identities.

Parameters:

Name	Type	Description	Default
threshold	float	Similarity threshold for matching nodes (0.0 to 1.0). Defaults to 0.8.	0.8

Returns:

Type	Description
iText2KG_Star	The updated iText2KG_Star instance with matched nodes and updated edges.

---

KG_Gen

hyperextract.methods.typical.KG_Gen

Bases: AutoGraph[NodeSchema, EdgeSchema]

Knowledge Graph Generator: A specialized AutoGraph for extracting simple triple-based KGs.

Features: - Fixed Schema (NodeSchema, EdgeSchema) optimized for triple extraction - Customized prompts from original kg_gen implementation - Automatic deduplication and consistency checking via AutoGraph's OMem - Two-stage extraction: entities first, then relationships

Example

llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) embedder = OpenAIEmbeddings() kg = KG_Gen(llm_client=llm, embedder=embedder) kg.feed_text(text) print(f"Extracted {len(kg.nodes)} entities and {len(kg.edges)} relationships.")

Functions

__init__(llm_client: BaseChatModel, embedder: Embeddings, chunk_size: int = 2048, chunk_overlap: int = 256, max_workers: int = 10, verbose: bool = False)

Initialize KGGenGraph.

Parameters:

Name	Type	Description	Default
llm_client	BaseChatModel	Language model for extraction	required
embedder	Embeddings	Embedding model for vector indexing	required
chunk_size	int	Characters per chunk	2048
chunk_overlap	int	Overlapping characters between chunks	256
max_workers	int	Max concurrent extraction workers	10
verbose	bool	Display detailed execution logs and progress information	False

_deduplicate_graph(graph_data: AutoGraphSchema[NodeSchema, EdgeSchema], threshold: float = 0.9) -> AutoGraphSchema[NodeSchema, EdgeSchema]

Internal helper to apply SemHash deduplication on a graph data object.

Parameters:

Name	Type	Description	Default
graph_data	AutoGraphSchema[NodeSchema, EdgeSchema]	The graph data object (nodes/edges) to process in-place.	required
threshold	float	SemHash similarity threshold (0.0 to 1.0).	0.9

Returns:

Type	Description
AutoGraphSchema[NodeSchema, EdgeSchema]	The modified graph_data object.

deduplicate(threshold: float = 0.9) -> KG_Gen

Return a NEW KG_Gen instance with deduplicated entities and edges. Does not modify the current instance.

Parameters:

Name	Type	Description	Default
threshold	float	Similarity threshold for SemHash (0.0 to 1.0). Higher means stricter.	0.9

Returns:

Type	Description
KG_Gen	A new, deduplicated KG_Gen instance.

self_deduplicate(threshold: float = 0.9) -> KG_Gen

Deduplicate the current graph IN-PLACE using SemHash.

Parameters:

Name	Type	Description	Default
threshold	float	Similarity threshold (0.0 to 1.0).	0.9

Returns:

Type	Description
KG_Gen	self (modified in-place)

---

Atom

hyperextract.methods.typical.Atom

Bases: AutoGraph[NodeSchema, EdgeSchema]

Atom: A specialized AutoGraph for extracting high-quality triple-based KGs.

Features: - Two-stage extraction: Extracts atomic facts first, then derives edges and nodes from facts - Customized prompts from original Atom implementation - Semantic Deduplication: Includes match_nodes_and_update_edges using SemHash with embeddings - Temporal tracking: t_start, t_end, and t_obs fields capture relationship timing and extraction metadata - Evidence attribution: atomic_facts field traces each extracted edge to source facts - Nested node schema: Maintains rich semantic information (name + label)

Example

from langchain_openai import ChatOpenAI, OpenAIEmbeddings llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) embedder = OpenAIEmbeddings() kg = Atom(llm_client=llm, embedder=embedder)

1. Extract: Facts -> Edges -> Nodes

kg.feed_text("Elon Musk is the CEO of SpaceX. Musk leads Tesla.") print(f"Before dedup - Nodes: {len(kg.nodes)}, Edges: {len(kg.edges)}")

2. Deduplicate: Merges 'Elon Musk' and 'Musk'

kg.match_nodes_and_update_edges(threshold=0.85) print(f"After dedup - Nodes: {len(kg.nodes)}, Edges: {len(kg.edges)}")

Attributes

facts_per_chunk = facts_per_chunk instance-attribute

observation_time = observation_time instance-attribute

Functions

__init__(llm_client: BaseChatModel, embedder: Embeddings, observation_time: str | None = None, chunk_size: int = 2048, chunk_overlap: int = 256, facts_per_chunk: int = 10, max_workers: int = 10, verbose: bool = False)

Initialize Atom.

Parameters:

Name	Type	Description	Default
llm_client	BaseChatModel	Language model for extraction	required
embedder	Embeddings	Embedding model for vector indexing	required
observation_time	str | None	Date when the extraction was performed, like '1997-10-10' or '1997-10-10 23:59:59'. If None, uses current date and time.	None
chunk_size	int	Characters per chunk	2048
chunk_overlap	int	Overlapping characters between chunks	256
facts_per_chunk	int	Max number of atomic facts to group into a single extraction batch (default: 10)	10
max_workers	int	Max concurrent extraction workers	10
verbose	bool	Display detailed execution logs and progress information	False

_extract_data(text: str) -> AutoGraphSchema

Extract atomic facts first, then extract edges (Two-stage extraction).

Process: 1. Split text into chunks. 2. Batch extract atomic facts from chunks. 3. Consolidate facts into a unified context. 4. Split consolidated facts into chunks. 5. Batch extract edges from fact chunks. 6. Post-process: Set t_obs timestamp & Derive nodes. 7. Merge all partial graphs into one global graph.

Parameters:

Name	Type	Description	Default
text	str	Input text.	required

Returns:

Type	Description
AutoGraphSchema	Extracted and validated graph.

match_nodes_and_update_edges(threshold: float = 0.8) -> Atom

Match nodes in the graph and update edges accordingly using SemHash with embeddings.

This method identifies and merges similar nodes based on semantic similarity using embeddings from the instance's embedder. It updates edges to reflect any changes in node identities.

Parameters:

Name	Type	Description	Default
threshold	float	Similarity threshold for matching nodes (0.0 to 1.0). Defaults to 0.8.	0.8

Returns:

Type	Description
Atom	The updated Atom instance with matched nodes and updated edges.