The Weaviate_VectorStores node is a component for integrating Weaviate, a scalable open-source vector database, into a vector store workflow. It allows for upsert operations of embedded data and performing similarity or MMR (Maximal Marginal Relevance) searches.
Name: Weaviate
Type: Weaviate
Version: 3.0
Category: Vector Stores
Weaviate
VectorStoreRetriever
BaseRetriever
Embeddings: The embedding model to use for vectorizing the data.
Weaviate Scheme: The scheme to use for connecting to Weaviate (https or http).
Weaviate Host: The host address of the Weaviate instance (e.g., localhost:8080).
Weaviate Index: The name of the Weaviate index to use.
Document: List of documents to be inserted into the vector store.
Record Manager: Keeps track of records to prevent duplication.
Weaviate Text Key: The key used for storing text data in Weaviate.
Weaviate Metadata Keys: JSON array of metadata keys to be stored in Weaviate.
Top K: Number of top results to fetch (default: 4).
Weaviate Search Filter: JSON object for filtering search results.
MMR Parameters: Additional parameters for Maximal Marginal Relevance search.
Weaviate Retriever: A retriever object for querying the Weaviate vector store.
Weaviate Vector Store: The Weaviate vector store object.
Upsert: Allows inserting or updating documents in the Weaviate vector store.
Delete: Enables deletion of documents from the vector store based on their IDs.
Initialization: Sets up the Weaviate client and creates a vector store instance.
Semantic search applications
Recommendation systems
Document similarity analysis
Knowledge management systems
Any application requiring efficient storage and retrieval of vector embeddings
The node supports both local and cloud-hosted Weaviate instances.
It integrates with a record manager for tracking and preventing duplicate entries.
The node allows for customization of text and metadata keys for flexible data storage.
It supports advanced querying capabilities through Weaviate’s filtering mechanism.