Retrieval and RAG Patterns
Auto-generated from data/terms.json. Do not edit this file manually.
Embedding
Embeddings are vector representations of data such as text, images or audio, where each item is mapped to a point in a continuous numerical space. The distance between points reflects semantic similarity, allowing machine learning systems to search, compare, cluster and classify information based on meaning. Embeddings are fundamental for tasks like semantic search, recommendations, anomaly detection and grouping related content.
Retrieval-Augmented Generation
Retrieval-Augmented Generation (RAG) is a technique where a model retrieves relevant information from an external knowledge source and uses it as context before generating a response. This improves the accuracy, grounding and reliability of the model by ensuring that answers are informed by up-to-date and authoritative data beyond the model's original training corpus.
Vector Database
A vector database is a database designed to store, manage, and index high-dimensional vector embeddings for efficient similarity search. It represents data as numerical vectors, clusters them based on semantic relatedness, and enables fast retrieval through specialized indexes. Vector databases often support metadata filtering, CRUD operations, and horizontal scaling, making them well suited for AI applications that require low-latency search across large datasets.