Vectorstore with BigQuery
BigQuery is Google Cloud’s fully managed, petabyte-scale data warehouse designed to handle the most demanding big data analytics workloads.
Expanding beyond structured data analytics, BigQuery’s recently introduced Vector Search capability allows for efficient similarity-based search on unstructured data. This functionality is seamlessly integrated into BigQuery, utilizing the same scalable infrastructure and familiar SQL interface.
BigQuery natively supports storing vector embeddings as arrays of floats. Vector search within BigQuery leverages the VECTOR_SEARCH function, optionally utilizing a vector index for enhanced performance.
When a vector index is employed, the VECTOR_SEARCH function utilizes an approximate nearest neighbor (ANN) search, improving search speed at the potential cost of slightly reduced recall. Conversely, when a vector index is not available or is explicitly disabled, BigQuery defaults to brute-force search for...
