Query NebulaGraph Knowledge Graphs with LlamaIndex

NebulaGraph Query Engine Pack

This LlamaPack creates a NebulaGraph query engine, and executes its query function. This pack offers the option of creating multiple types of query engines, namely:

• Knowledge graph vector-based entity retrieval (default if no query engine type option is provided)
• Knowledge graph keyword-based entity retrieval
• Knowledge graph hybrid entity retrieval
• Raw vector index retrieval
• Custom combo query engine (vector similarity + KG entity retrieval)
• KnowledgeGraphQueryEngine
• KnowledgeGraphRAGRetriever

CLI Usage

You can download llamapacks directly using llamaindex-cli, which comes installed with the llama-index python package:

llamaindex-cli download-llamapack NebulaGraphQueryEnginePack --download-dir ./nebulagraph_pack

You can then inspect the files at ./nebulagraph_pack and use them as a template for your own project!

Code Usage

You can download the pack to a ./nebulagraph_pack directory:

from llama_index.core.llama_pack import download_llama_pack

### download and install dependencies
NebulaGraphQueryEnginePack = download_llama_pack(
    "NebulaGraphQueryEnginePack", "./nebulagraph_pack"
)

From here, you can use the pack, or inspect and modify the pack in ./nebulagraph_pack.

Then, you can set up the pack like so:

pip install llama-index-readers-wikipedia

### Load the docs (example of Paleo diet from Wikipedia)

from llama_index.readers.wikipedia import WikipediaReader

loader = WikipediaReader()
docs = loader.load_data(pages=["Paleolithic diet"], auto_suggest=False)
print(f"Loaded {len(docs)} documents")

### get NebulaGraph credentials (assume it's stored in credentials.json)
with open("credentials.json") as f:
    nebulagraph_connection_params = json.load(f)
    username = nebulagraph_connection_params["username"]
    password = nebulagraph_connection_params["password"]
    ip_and_port = nebulagraph_connection_params["ip_and_port"]

space_name = "paleo_diet"
edge_types, rel_prop_names = ["relationship"], ["relationship"]
tags = ["entity"]
max_triplets_per_chunk = 10

### create the pack
nebulagraph_pack = NebulaGraphQueryEnginePack(
    username=username,
    password=password,
    ip_and_port=ip_and_port,
    space_name=space_name,
    edge_types=edge_types,
    rel_prop_names=rel_prop_names,
    tags=tags,
    max_triplets_per_chunk=max_triplets_per_chunk,
    docs=docs,
)

Optionally, you can pass in the query_engine_type from NebulaGraphQueryEngineType to construct NebulaGraphQueryEnginePack. If query_engine_type is not defined, it defaults to Knowledge Graph vector based entity retrieval.

from llama_index.core.packs.nebulagraph_query_engine.base import (
    NebulaGraphQueryEngineType,
)

### create the pack
nebulagraph_pack = NebulaGraphQueryEnginePack(
    username=username,
    password=password,
    ip_and_port=ip_and_port,
    space_name=space_name,
    edge_types=edge_types,
    rel_prop_names=rel_prop_names,
    tags=tags,
    max_triplets_per_chunk=max_triplets_per_chunk,
    docs=docs,
    query_engine_type=NebulaGraphQueryEngineType.KG_HYBRID,
)

NebulaGraphQueryEnginePack is a enum defined as follows:

class NebulaGraphQueryEngineType(str, Enum):
    """NebulaGraph query engine type"""

    KG_KEYWORD = "keyword"
    KG_HYBRID = "hybrid"
    RAW_VECTOR = "vector"
    RAW_VECTOR_KG_COMBO = "vector_kg"
    KG_QE = "KnowledgeGraphQueryEngine"
    KG_RAG_RETRIEVER = "KnowledgeGraphRAGRetriever"

The run() function is a light wrapper around query_engine.query(), see a sample query below.

response = nebulagraph_pack.run("Tell me about the benefits of paleo diet.")

You can also use modules individually.

### call the query_engine.query()
query_engine = nebulagraph_pack.query_engine
response = query_engine.query("query_str")