Definition
An integrated vector database capability within MongoDB Atlas that enables the storage, indexing, and querying of high-dimensional embeddings alongside structured document data. In RAG pipelines, it facilitates semantic search by utilizing the k-Nearest Neighbors (k-NN) algorithm to retrieve contextually relevant data segments based on vector proximity.
Distinguish from MongoDB's legacy $text search; Vector Search uses mathematical distance in latent space rather than keyword frequency.
"A warehouse where items are stored in a single room, but are found using a GPS coordinate (vector) while simultaneously checking the item's barcode and expiration date (metadata)."
- Approximate Nearest Neighbor (ANN)(Underlying Search Algorithm)
- Hybrid Search(Query Strategy)
- Cosine Similarity(Distance Metric)
- Metadata Filtering(Component)
Conceptual Overview
An integrated vector database capability within MongoDB Atlas that enables the storage, indexing, and querying of high-dimensional embeddings alongside structured document data. In RAG pipelines, it facilitates semantic search by utilizing the k-Nearest Neighbors (k-NN) algorithm to retrieve contextually relevant data segments based on vector proximity.
Disambiguation
Distinguish from MongoDB's legacy $text search; Vector Search uses mathematical distance in latent space rather than keyword frequency.
Visual Analog
A warehouse where items are stored in a single room, but are found using a GPS coordinate (vector) while simultaneously checking the item's barcode and expiration date (metadata).