向量存储集成

使用 LangChain JavaScript 集成向量存储

概述

一个 向量存储 存储了 嵌入 数据，并执行相似性搜索。

接口

LangChain 提供了一个统一的向量存储接口，允许你：

• addDocuments - 向存储中添加文档。
• delete - 通过 ID 删除存储的文档。
• similaritySearch - 查询语义相似的文档。

这种抽象让你可以在不同的实现之间切换，而无需更改应用程序逻辑。

初始化

LangChain 中的大多数向量存储在初始化向量存储时接受一个嵌入模型作为参数。

import { OpenAIEmbeddings } from "@langchain/openai";
import { MemoryVectorStore } from "@langchain/classic/vectorstores/memory";

const embeddings = new OpenAIEmbeddings({
  model: "text-embedding-3-small",
});
const vectorStore = new MemoryVectorStore(embeddings);

添加文档

你可以通过使用 addDocuments 函数向向量存储中添加文档。

import { Document } from "@langchain/core/documents";
const document = new Document({
  pageContent: "Hello world",
});
await vectorStore.addDocuments([document]);

删除文档

你可以通过使用 delete 函数从向量存储中删除文档。

await vectorStore.delete({
  filter: {
    pageContent: "Hello world",
  },
});

相似性搜索

使用 similaritySearch 发起语义查询，它将返回最接近的嵌入文档：

const results = await vectorStore.similaritySearch("Hello world", 10);

许多向量存储支持以下参数：

• k — 返回的结果数量
• filter — 基于元数据的条件过滤

相似性指标与索引

嵌入相似性可能使用以下方法计算：

• 余弦相似度
• 欧几里得距离
• 点积

高效的搜索通常采用 HNSW（分层可导航小世界）等索引方法，但具体细节取决于向量存储。

元数据过滤

通过元数据（例如，来源、日期）过滤可以细化搜索结果：

vectorStore.similaritySearch("query", 2, { source: "tweets" });

顶级集成

选择嵌入模型：

OpenAI

安装依赖项：

npm

npm install @langchain/openai @langchain/core

添加环境变量：

OPENAI_API_KEY=your-api-key

实例化模型：

import { OpenAIEmbeddings } from "@langchain/openai";

const embeddings = new OpenAIEmbeddings({
  model: "text-embedding-3-large"
});

Azure

安装依赖项

npm

npm install @langchain/openai @langchain/core

添加环境变量：

AZURE_OPENAI_API_INSTANCE_NAME=<YOUR_INSTANCE_NAME>
AZURE_OPENAI_API_KEY=<YOUR_KEY>
AZURE_OPENAI_API_VERSION="2024-02-01"

实例化模型：

import { AzureOpenAIEmbeddings } from "@langchain/openai";

const embeddings = new AzureOpenAIEmbeddings({
  azureOpenAIApiEmbeddingsDeploymentName: "text-embedding-ada-002"
});

AWS

安装依赖项：

npm

npm i @langchain/aws

添加环境变量：

BEDROCK_AWS_REGION=your-region

实例化模型：

import { BedrockEmbeddings } from "@langchain/aws";

const embeddings = new BedrockEmbeddings({
  model: "amazon.titan-embed-text-v1"
});

Google Gemini

安装依赖项：

npm

npm i @langchain/google-genai

添加环境变量：

GOOGLE_API_KEY=your-api-key

实例化模型：

import { GoogleGenerativeAIEmbeddings } from "@langchain/google-genai";

const embeddings = new GoogleGenerativeAIEmbeddings({
  model: "gemini-embedding-001"
});

Gemini Enterprise Agent Platform

安装依赖项：

npm

npm i @langchain/google-vertexai

添加环境变量：

GOOGLE_APPLICATION_CREDENTIALS=credentials.json

实例化模型：

import { VertexAIEmbeddings } from "@langchain/google-vertexai";

const embeddings = new VertexAIEmbeddings({
  model: "gemini-embedding-001"
});

MistralAI

安装依赖：

npm

npm install @langchain/mistralai @langchain/core

添加环境变量：

MISTRAL_API_KEY=your-api-key

实例化模型：

import { MistralAIEmbeddings } from "@langchain/mistralai";

const embeddings = new MistralAIEmbeddings({
  model: "mistral-embed"
});

Cohere

安装依赖：

npm

npm i @langchain/cohere

添加环境变量：

COHERE_API_KEY=your-api-key

实例化模型：

import { CohereEmbeddings } from "@langchain/cohere";

const embeddings = new CohereEmbeddings({
  model: "embed-english-v3.0"
});

Ollama

安装依赖：

npm

npm install @langchain/ollama @langchain/core

实例化模型：

import { OllamaEmbeddings } from "@langchain/ollama";

const embeddings = new OllamaEmbeddings({
  model: "llama2",
  baseUrl: "http://localhost:11434", // Default value
});

Voyage AI

安装依赖：

npm

npm install @langchain/mongodb @langchain/core

添加环境变量：

VOYAGE_API_KEY=your-api-key

实例化模型：

import { VoyageEmbeddings } from "@langchain/mongodb";

const embeddings = new VoyageEmbeddings({
  model: "voyage-4"
});

选择向量存储：

Memory

npm i langchain

import { MemoryVectorStore } from "@langchain/classic/vectorstores/memory";

const vectorStore = new MemoryVectorStore(embeddings);

MongoDB

Manual embedding

npm

npm install @langchain/mongodb mongodb @langchain/core

import { MongoDBAtlasVectorSearch } from "@langchain/mongodb"
import { MongoClient } from "mongodb";

const client = new MongoClient(process.env.MONGODB_ATLAS_URI!);
const collection = client
  .db(process.env.MONGODB_ATLAS_DB_NAME)
  .collection(process.env.MONGODB_ATLAS_COLLECTION_NAME);

const vectorStore = new MongoDBAtlasVectorSearch(embeddings, {
  collection,
  indexName: "vector_index",
  textKey: "text",
  embeddingKey: "embedding",
});

Automated embedding

npm

npm install @langchain/mongodb mongodb @langchain/core

import { MongoDBAtlasVectorSearch } from "@langchain/mongodb"
import { MongoClient } from "mongodb";

const client = new MongoClient(process.env.MONGODB_ATLAS_URI!);
const collection = client
  .db(process.env.MONGODB_ATLAS_DB_NAME)
  .collection(process.env.MONGODB_ATLAS_COLLECTION_NAME);

const vectorStore = new MongoDBAtlasVectorSearch({ collection });

Pinecone

npm

npm install @langchain/pinecone @langchain/core @pinecone-database/pinecone

import { PineconeStore } from "@langchain/pinecone";
import { Pinecone as PineconeClient } from "@pinecone-database/pinecone";

const pinecone = new PineconeClient();
const vectorStore = new PineconeStore(embeddings, {
  pineconeIndex,
  maxConcurrency: 5,
});

Redis

npm

npm install @langchain/redis @langchain/core redis

import { RedisVectorStore } from "@langchain/redis";

const vectorStore = new RedisVectorStore(embeddings, {
  redisClient: client,
  indexName: "langchainjs-testing",
});

Qdrant

npm

npm install @langchain/qdrant @langchain/core

import { QdrantVectorStore } from "@langchain/qdrant";

const vectorStore = await QdrantVectorStore.fromExistingCollection(embeddings, {
  url: process.env.QDRANT_URL,
  collectionName: "langchainjs-testing",
});

Oracle AI Database

npm

npm i @oracle/langchain-oracledb @langchain/core

import oracledb from "oracledb";
import { OracleEmbeddings, OracleVS } from "@oracle/langchain-oracledb";

const connection = await oracledb.getConnection({
  user: process.env.ORACLE_USER,
  password: process.env.ORACLE_PASSWORD,
  connectionString: process.env.ORACLE_DSN,
});

const embeddings = new OracleEmbeddings(connection, {
  provider: "database",
  model: process.env.DEMO_ONNX_MODEL ?? "DEMO_MODEL",
});

const vectorStore = new OracleVS(embeddings, {
  client: connection,
  tableName: "DEMO_VECTORS",
  query: "Find support tickets mentioning service outages.",
  distanceStrategy: "DOT",
});
await vectorStore.initialize();

Weaviate

npm

npm install @langchain/weaviate @langchain/core weaviate-client

import { WeaviateStore } from "@langchain/weaviate";

const vectorStore = new WeaviateStore(embeddings, {
    client: weaviateClient,
    indexName: "Langchainjs_test",
});

LangChain.js 与多种向量存储集成。以下为完整列表：

所有向量存储

向量存储	下载
WeaviateStore
PineconeStore
MongoDBAtlasVectorSearch
QdrantVectorStore
RedisVectorStore
OracleVS
PGVectorStore
Cloudflare vectorize
TurbopufferVectorStore
Google cloud SQL for postgresql
Azure Cosmos DB for MongoDB vCore (deprecated)
Azure Cosmos DB for NoSQL
Azure DocumentDB
Neo4jVectorStore
SAP HANA Cloud Vector Engine
LambdaDB
Infino
YDB
langchain	N/A

---

官方原文：src/oss/javascript/integrations/vectorstores/index.mdx · 非官方翻译，译文由 AI 生成，以英文原文为准。