retrieve_graph

Retrieves and returns related document chunks.

def retrieve_graph(question, **kwargs)

Parameter	Data Type	Required/Optional	Description
question	String	Required	User question. The value is a string of 1 to 1000 × 1000 characters.
kwargs	Dict	Optional	Extended parameters: use_text (bool): The default value is True, indicating that only the text is used to construct the context during subgraph retrieval. batch_size (integer): batch size when nodes are vectorized. The value range is [1, 1024] and the default value is 4. similarity_tail_threshold: vector similarity threshold. The default value is 0.0. Similarity results below this value will be filtered. The value range is [0.0, 1.0]. retrieval_top_k (integer): topk results when similar nodes are retrieved from the node vector database based on entities. The value range is [1, 1000] and the default value is 40. reranker_top_k: topk results required by the reranker. The value range is [1, 1000], and the default value is 20. subgraph_depth (integer): maximum depth of graph retrieval. The value range is [1, 5] and the default value is 2.

Data Type	Description
List[str]	Retrieved context chunks

Parent topic: GraphRAGPipeline