참고소스 수정본

참고/neo4j-graphrag-python-main/examples/customize/build_graph/components/extractors/custom_extractor.py

@@ -0,0 +1,41 @@
+from typing import Any, Optional
+
+from neo4j_graphrag.experimental.components.entity_relation_extractor import (
+    EntityRelationExtractor,
+    OnError,
+)
+from neo4j_graphrag.experimental.components.types import (
+    DocumentInfo,
+    LexicalGraphConfig,
+    Neo4jGraph,
+    TextChunks,
+)
+
+
+class MyExtractor(EntityRelationExtractor):
+    def __init__(
+        self,
+        *args: Any,
+        on_error: OnError = OnError.IGNORE,
+        create_lexical_graph: bool = True,
+        **kwargs: Any,
+    ) -> None:
+        super().__init__(
+            *args,
+            on_error=on_error,
+            create_lexical_graph=create_lexical_graph,
+            **kwargs,
+        )
+
+    async def run(
+        self,
+        chunks: TextChunks,
+        document_info: Optional[DocumentInfo] = None,
+        lexical_graph_config: Optional[LexicalGraphConfig] = None,
+        **kwargs: Any,
+    ) -> Neo4jGraph:
+        # Implement your logic here
+        # you can loop over all text chunks with:
+        for chunk in chunks.chunks:
+            pass
+        return Neo4jGraph(nodes=[], relationships=[])

참고/neo4j-graphrag-python-main/examples/customize/build_graph/components/extractors/llm_entity_relation_extractor.py

@@ -0,0 +1,33 @@
+from neo4j_graphrag.experimental.components.entity_relation_extractor import (
+    LLMEntityRelationExtractor,
+)
+from neo4j_graphrag.experimental.components.types import (
+    Neo4jGraph,
+    TextChunk,
+    TextChunks,
+)
+from neo4j_graphrag.llm import LLMInterface
+
+
+async def main(llm: LLMInterface) -> Neo4jGraph:
+    """
+
+    Args:
+        llm (LLMInterface): Any LLM implemented in neo4j_graphrag.llm or from LangChain chat models.
+    """
+    extractor = LLMEntityRelationExtractor(
+        llm=llm,
+        # optional: customize the prompt used for entity and relation extraction
+        # prompt_template="",
+        # optional: disable the creation of the lexical graph (Document and Chunk nodes)
+        # create_lexical_graph=False,
+        # optional: if an LLM error happens, ignore the chunk and continue process with the next ones
+        # default value is OnError.RAISE which will end the process
+        # on_error=OnError.IGNORE,
+        # optional: tune the max_concurrency parameter to optimize speed
+        # max_concurrency=5,
+    )
+    graph = await extractor.run(
+        chunks=TextChunks(chunks=[TextChunk(text="....", index=0)])
+    )
+    return graph

참고/neo4j-graphrag-python-main/examples/customize/build_graph/components/extractors/llm_entity_relation_extractor_with_custom_prompt.py

@@ -0,0 +1,33 @@
+from neo4j_graphrag.experimental.components.entity_relation_extractor import (
+    LLMEntityRelationExtractor,
+)
+from neo4j_graphrag.experimental.components.types import (
+    Neo4jGraph,
+    TextChunk,
+    TextChunks,
+)
+from neo4j_graphrag.llm import LLMInterface
+
+
+async def main(llm: LLMInterface) -> Neo4jGraph:
+    """
+
+    Args:
+        llm (LLMInterface): Any LLM implemented in neo4j_graphrag.llm or from LangChain chat models.
+    """
+    extractor = LLMEntityRelationExtractor(
+        llm=llm,
+        # optional: customize the prompt used for entity and relation extraction
+        # prompt_template="",
+        # optional: disable the creation of the lexical graph (Document and Chunk nodes)
+        # create_lexical_graph=False,
+        # optional: if an LLM error happens, ignore the chunk and continue process with the next ones
+        # default value is OnError.RAISE which will end the process
+        # on_error=OnError.IGNORE,
+        # optional: tune the max_concurrency parameter to optimize speed
+        # max_concurrency=5,
+    )
+    graph = await extractor.run(
+        chunks=TextChunks(chunks=[TextChunk(text="....", index=0)])
+    )
+    return graph

참고/neo4j-graphrag-python-main/examples/customize/build_graph/components/extractors/llm_entity_relation_extractor_with_structured_output.py

@@ -0,0 +1,74 @@
+"""
+Simple example demonstrating structured output with LLMEntityRelationExtractor.
+
+This example shows how to use structured output for more reliable entity and
+relationship extraction with automatic schema validation.
+
+The Neo4jGraph schema is now compatible with both OpenAI and VertexAI structured
+output APIs, with strict schema validation (additionalProperties: false) and
+proper required field definitions.
+
+Prerequisites:
+- Google Cloud credentials configured for VertexAI
+- Or OpenAI API key set in OPENAI_API_KEY environment variable
+"""
+
+import asyncio
+from dotenv import load_dotenv
+
+from neo4j_graphrag.experimental.components.entity_relation_extractor import (
+    LLMEntityRelationExtractor,
+)
+from neo4j_graphrag.experimental.components.types import (
+    Neo4jGraph,
+    TextChunk,
+    TextChunks,
+)
+from neo4j_graphrag.llm import VertexAILLM
+
+
+async def main() -> Neo4jGraph:
+    """
+    Demonstrates entity and relation extraction with structured output.
+
+    With use_structured_output=True:
+    - Uses LLMInterfaceV2 (list of messages)
+    - Passes Neo4jGraph Pydantic model as response_format to invoke()
+    - Ensures response conforms to expected graph structure
+    - Provides automatic type validation
+    - Reduces need for JSON repair and error handling
+    """
+    load_dotenv()
+    # Initialize LLM - no response_format in constructor!
+    llm = VertexAILLM(model_name="gemini-2.5-flash")
+
+    # llm = OpenAILLM(
+    #     model_name="gpt-5-mini",
+    #     model_params={"temperature": 0}
+    # )
+
+    # Enable structured output for reliable extraction
+    extractor = LLMEntityRelationExtractor(
+        llm=llm,
+        use_structured_output=True,  # This is the key parameter!
+    )
+
+    # Sample text about a person and organization
+    sample_text = """
+    Albert Einstein was a theoretical physicist who developed the theory of relativity.
+    He worked at the Institute for Advanced Study in Princeton from 1933 until his death in 1955.
+    """
+
+    # Extract entities and relationships
+    graph = await extractor.run(
+        chunks=TextChunks(chunks=[TextChunk(text=sample_text, index=0)])
+    )
+
+    return graph
+
+
+if __name__ == "__main__":
+    # Run extraction
+    graph = asyncio.run(main())
+
+    print(graph)