· Embedding: We transform raw data (text, images) into vectors using machine learning techniques. Imagine a complex algorithm summarising a book into a series of key points.
· Indexing: The vectors are then stored and meticulously indexed within the database. This is where the magic happens – the database structures the data for efficient retrieval.
- Query engine: The query engine facilitates fast and accurate similarity search within the vector space. It employs algorithms like Approximate Nearest Neighbor (ANN) search techniques, leveraging indexing structures to efficiently locate vectors that are close in proximity to a given query vector.