You store embeddings in a property called embedding within an Azure Cosmos DB for NoSQL container.
embedding
You must retrieve the five documents most similar to an input embedding.
You need to build the similarity query. What should you do?
Azure Cosmos DB for NoSQL vector queries order items by VectorDistance and use TOP N to return the required number of nearest matches. A TOP 5 clause limits the result set to the five most similar documents; Microsoft recommends using TOP N for vector searches to avoid returning unnecessary results.
VectorDistance
TOP N
TOP 5
Community Discussion