In vector search, what does the term 'query vector' refer to?

The term 'query vector' specifically refers to a vector that represents the input for which similar items are being searched. In the context of vector search, a query vector is created by transforming the input data (such as text, images, or other forms of information) into a multi-dimensional numerical representation. This vector serves as a reference point in the vector space, allowing the search algorithm to identify and retrieve other vectors (representing items in the dataset) that are most similar to the query vector based on specific similarity metrics, such as cosine similarity or Euclidean distance.

By using the query vector to perform searches, systems can efficiently find and recommend relevant items, making it a fundamental concept in vector search applications. Other choices, like vectors representing multiple data points or designed for storage efficiency, do not capture the specific function of a query vector in the search process.