Meilisearch

​

Node Details

• Name: meilisearch
• Type: Meilisearch
• Version: 1.0
• Category: Vector Stores

​

Base Classes

• BaseRetriever

​

Inputs

1. Document (optional, list)
   • Type: Document
   • Description: List of documents to be embedded and stored
2. Embeddings
   • Type: Embeddings
   • Description: Embedding model to use for vectorizing documents
3. Host
   • Type: string
   • Description: URL for the desired Meilisearch instance (must not end with a ’/’)
4. Index Uid
   • Type: string
   • Description: UID for the index to answer from
5. Delete Index if exists (optional)
   • Type: boolean
   • Description: Whether to delete the existing index before upserting
6. Top K (optional, additional param)
   • Type: number
   • Description: Number of top searches to return as context (default is 4)
7. Semantic Ratio (optional, additional param)
   • Type: number
   • Description: Percentage of semantic reasoning in Meilisearch hybrid search (default is 0.75)
8. Search Filter (optional, additional param)
   • Type: string
   • Description: Search filter to apply on searchable attributes

​

Outputs

• Meilisearch Retriever
  • Name: retriever
  • Base Classes: [BaseRetriever]
  • Description: Retriever for getting answers from the Meilisearch index

​

Credentials

• Credential Name: meilisearchApi
• Parameters:
  • meilisearchAdminApiKey
  • meilisearchSearchApiKey

​

Functionality

1. Upsert:
   • Embeds documents using the provided embedding model
   • Creates or updates a Meilisearch index with the embedded documents
   • Handles index deletion if requested
   • Sets up necessary index settings for vector search
2. Initialization:
   • Sets up the Meilisearch client with the provided credentials
   • Enables vector store features in Meilisearch
   • Creates a MeilisearchRetriever instance for similarity search

​

Usage

• Creating and managing document embeddings in a Meilisearch index
• Performing hybrid (keyword + semantic) searches on stored documents
• Integrating Meilisearch’s vector search capabilities into a larger workflow