Perfrom LocalSearch withouth combining with embeding store.

[jorgelunams]

How can we run LocalSearch withouth

entity_df = pd.read_parquet(f"{INPUT_DIR}/{ENTITY_TABLE}.parquet")
entity_embedding_df = pd.read_parquet(f"{INPUT_DIR}/{ENTITY_EMBEDDING_TABLE}.parquet")

entities = read_indexer_entities(entity_df, entity_embedding_df, COMMUNITY_LEVEL)

load description embeddings to an in-memory lancedb vectorstore

to connect to a remote db, specify url and port values.

description_embedding_store = LanceDBVectorStore(
collection_name="entity_description_embeddings",
)
description_embedding_store.connect(db_uri=LANCEDB_URI)
entity_description_embeddings = store_entity_semantic_embeddings(
entities=entities, vectorstore=description_embedding_store
)
print(LANCEDB_URI)
print(f"Entity count: {len(entity_df)}")
entity_df.head()

In our case we haave only the Knowledge Graph and not access to a Vector DB like the search sample for operation dulce.

[natoverse]

You could do this with a bit of work. The CLI can be a helpful example of how to load and execute a search engine. In this code, we invoke a factory function that creates an instance of the LocalSearch class. This factory function accepts a description_embedding_store parameter which needs to be an implementation of BaseVectorStore. We supply implementations for Azure AI Search and LanceDB in GraphRAG, but you could write your own that does whatever lookup you need to supply the entity matches.