安装带有 `astra` 额外依赖的 `langchain-graph-retriever` 包：
```bash theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
pip install "langchain-graph-retriever[astra]"
```
然后创建一个向量存储并加载测试文档：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_astradb import AstraDBVectorStore
vector_store = AstraDBVectorStore.from_documents(
documents=animals,
embedding=embeddings,
collection_name="animals",
api_endpoint=ASTRA_DB_API_ENDPOINT,
token=ASTRA_DB_APPLICATION_TOKEN,
)
```
有关 `ASTRA_DB_API_ENDPOINT` 和 `ASTRA_DB_APPLICATION_TOKEN` 凭据，请参阅 [AstraDB 向量存储指南](/oss/python/integrations/vectorstores/astradb)。
:::note
为了更快的初始测试，考虑使用 **InMemory** 向量存储。
:::
安装带有 `cassandra` 额外依赖的 `langchain-graph-retriever` 包：
```bash theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
pip install "langchain-graph-retriever[cassandra]"
```
然后创建一个向量存储并加载测试文档：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_community.vectorstores.cassandra import Cassandra
from langchain_graph_retriever.transformers import ShreddingTransformer
vector_store = Cassandra.from_documents(
documents=list(ShreddingTransformer().transform_documents(animals)),
embedding=embeddings,
table_name="animals",
)
```
有关创建 Cassandra 连接的帮助，请参阅
[Apache Cassandra 向量存储指南](/oss/python/integrations/vectorstores/cassandra#connection-parameters)
:::note
Apache Cassandra 不支持在嵌套元数据中搜索。因此，在插入文档时需要使用 [`ShreddingTransformer`](https://datastax.github.io/graph-rag/reference/langchain_graph_retriever/transformers/#langchain_graph_retriever.transformers.shredding.ShreddingTransformer)。
:::
安装带有 `opensearch` 额外依赖的 `langchain-graph-retriever` 包：
```bash theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
pip install "langchain-graph-retriever[opensearch]"
```
然后创建一个向量存储并加载测试文档：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_community.vectorstores import OpenSearchVectorSearch
vector_store = OpenSearchVectorSearch.from_documents(
documents=animals,
embedding=embeddings,
engine="faiss",
index_name="animals",
opensearch_url=OPEN_SEARCH_URL,
bulk_size=500,
)
```
有关创建 OpenSearch 连接的帮助，请参阅
[OpenSearch 向量存储指南](/oss/python/integrations/vectorstores/opensearch)。
安装带有 `chroma` 额外依赖的 `langchain-graph-retriever` 包：
```bash theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
pip install "langchain-graph-retriever[chroma]"
```
然后创建一个向量存储并加载测试文档：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_chroma.vectorstores import Chroma
from langchain_graph_retriever.transformers import ShreddingTransformer
vector_store = Chroma.from_documents(
documents=list(ShreddingTransformer().transform_documents(animals)),
embedding=embeddings,
collection_name="animals",
)
```
有关创建 Chroma 连接的帮助，请参阅
[Chroma 向量存储指南](/oss/python/integrations/vectorstores/chroma)。
:::note
Chroma 不支持在嵌套元数据中搜索。因此，在插入文档时需要使用 [`ShreddingTransformer`](https://datastax.github.io/graph-rag/reference/langchain_graph_retriever/transformers/#langchain_graph_retriever.transformers.shredding.ShreddingTransformer)。
:::
安装 `langchain-graph-retriever` 包：
```bash theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
pip install "langchain-graph-retriever"
```
然后创建一个向量存储并加载测试文档：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from langchain_core.vectorstores import InMemoryVectorStore
vector_store = InMemoryVectorStore.from_documents(
documents=animals,
embedding=embeddings,
)
```
:::tip
使用 `InMemoryVectorStore` 是开始使用 Graph RAG 的最快方式，但不建议用于生产环境。建议使用 **AstraDB** 或 **OpenSearch**。
:::
### 图遍历
此图检索器从与查询最匹配的单个动物开始，然后遍历到共享相同 `habitat` 和/或 `origin` 的其他动物。
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
from graph_retriever.strategies import Eager
from langchain_graph_retriever import GraphRetriever
traversal_retriever = GraphRetriever(
store = vector_store,
edges = [("habitat", "habitat"), ("origin", "origin")],
strategy = Eager(k=5, start_k=1, max_depth=2),
)
```
上述代码创建了一个图遍历检索器，它从最近的动物（`start_k=1`）开始，检索 5 个文档（`k=5`），并将搜索限制在距离第一个动物最多 2 步的文档（`max_depth=2`）。
`edges` 定义了如何使用元数据值进行遍历。在这种情况下，每个动物都与具有相同 `habitat` 和/或 `origin` 的其他动物相连。
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
results = traversal_retriever.invoke("what animals could be found near a capybara?")
for doc in results:
print(f"{doc.id}: {doc.page_content}")
```
```text theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
capybara: capybaras are the largest rodents in the world and are highly social animals.
heron: herons are wading birds known for their long legs and necks, often seen near water.
crocodile: crocodiles are large reptiles with powerful jaws and a long lifespan, often living over 70 years.
frog: frogs are amphibians known for their jumping ability and croaking sounds.
duck: ducks are waterfowl birds known for their webbed feet and quacking sounds.
```
图遍历通过利用数据中的结构化关系来提高检索质量。与标准相似性搜索（见下文）不同，它为文档的选择提供了清晰、可解释的理由。
在这种情况下，文档 `capybara`、`heron`、`frog`、`crocodile` 和 `newt` 都共享相同的 `habitat=wetlands`，如其元数据所定义。这应该会增加文档相关性并提高 LLM 答案的质量。
### 与标准检索的比较
当 `max_depth=0` 时，图遍历检索器的行为类似于标准检索器：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
standard_retriever = GraphRetriever(
store = vector_store,
edges = [("habitat", "habitat"), ("origin", "origin")],
strategy = Eager(k=5, start_k=5, max_depth=0),
)
```
这创建了一个从最近的 5 个动物（`start_k=5`）开始，并在没有任何遍历（`max_depth=0`）的情况下返回它们的检索器。在这种情况下，边定义被忽略。
这本质上等同于：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
standard_retriever = vector_store.as_retriever(search_kwargs={"k":5})
```
对于任一情况，调用检索器返回：
```python theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
results = standard_retriever.invoke("what animals could be found near a capybara?")
for doc in results:
print(f"{doc.id}: {doc.page_content}")
```
```text theme={"theme":{"light":"catppuccin-latte","dark":"catppuccin-mocha"}}
capybara: capybaras are the largest rodents in the world and are highly social animals.
iguana: iguanas are large herbivorous lizards often found basking in trees and near water.
guinea pig: guinea pigs are small rodents often kept as pets due to their gentle and social nature.
hippopotamus: hippopotamuses are large semi-aquatic mammals known for their massive size and territorial behavior.
boar: boars are wild relatives of pigs, known for their tough hides and tusks.
```
这些文档仅基于相似性进行连接。存储中存在的任何结构化数据都被忽略。与图检索相比，这可能会降低文档相关性，因为返回的结果对回答查询的帮助可能性较低。
## 用法
按照上面的示例，使用 `invoke` 来启动对查询的检索。
***
## API 参考
要探索所有可用参数和高级配置，请参阅
[Graph RAG API 参考](https://datastax.github.io/graph-rag/reference/)。
***