OpenSearch Vector Store

The OpenSearch Vector Store node is a component that integrates OpenSearch, an open-source, all-in-one vector database, into a vector store system. It allows for upsert operations of embedded data and similarity search queries.

Node Details

• Name: OpenSearch_VectorStores

• Type: OpenSearch

• Version: 3.0

• Category: Vector Stores

Base Classes

• OpenSearch

• VectorStoreRetriever

• BaseRetriever

Credentials

• Label: Connect Credential

• Name: credential

• Type: credential

• Credential Names: openSearchUrl

Input Parameters

1. Document

   • Type: Document

   • List: true

   • Optional: true

2. Embeddings

   • Type: Embeddings

3. Index Name

   • Type: string

4. Top K

   • Type: number

   • Description: Number of top results to fetch

   • Default: 4

   • Optional: true

Outputs

1. OpenSearch Retriever

   • Name: retriever

   • Base Classes: [OpenSearch, VectorStoreRetriever, BaseRetriever]

2. OpenSearch Vector Store

   • Name: vectorStore

   • Base Classes: [OpenSearch, …BaseClasses(OpenSearchVectorStore)]

Functionality

1. Upsert Method

   • Adds or updates documents in the OpenSearch index

   • Utilizes the provided embeddings to vectorize documents

   • Returns the number of added documents and the added documents themselves

2. Init Method

   • Initializes the OpenSearch vector store or retriever based on the specified output

   • Configures the client connection using provided credentials

   • Sets up the vector store with the given embeddings and index name

Usage

This node is used for:

1. Storing and indexing document embeddings in OpenSearch

2. Retrieving similar documents based on vector similarity

3. Integrating OpenSearch as a vector store in a larger language model or information retrieval system

OpenSearch Client Configuration

The node includes a utility function getOpenSearchClient that creates an OpenSearch client instance. It handles authentication if a username and password are provided.

Note

This node is part of a larger system and is designed to be used in conjunction with other nodes for building complex AI and information retrieval workflows.