Source
Unifying Structured and Unstructured Data Insights with BQ Search Innovations
Google has introduced updates to BigQuery for unifying structured and unstructured data insights. These include the general availability of Autonomous Embedding Generation and AI.SEARCH(), as well as a public preview of Hybrid Search.
Why it matters
These features reduce the need for complex external pipelines when analyzing documents like PDFs. This allows businesses to perform natural language searches over unstructured data more cost-effectively.
The details
- Autonomous Embedding Generation is generally available for text and image data.
- AI.SEARCH() now offers up to 133x gain in single-query slot efficiency.
- Hybrid Search combines semantic vector search with lexical keyword matching.
Show entities and relationshipsHide entities and relationships
In this article
Key connections
BigQuery supports external Vertex AI text-embedding models for autonomous embedding generation.
BigQuery natively utilizes Gemma embedding models for autonomous embedding generation.
BigQuery uses Google Cloud Storage
BigQuery queries unstructured documents stored in Google Cloud Storage using Object Tables.
BigQuery uses Retrieval-Augmented Generation
BigQuery simplifies RAG pipeline creation with autonomous embedding generation.
BigQuery is built with Hybrid Search
BigQuery introduces Hybrid Search combining semantic vector search and lexical search.
BigQuery is built with Vector Search
BigQuery integrates semantic vector search capabilities.
Show 5 more connectionsShow fewer connections
BigQuery uses Multimodal Search
BigQuery supports autonomous embedding generation over images using ObjectRefs for multimodal search.
Hybrid Search is built with Reciprocal Rank Fusion
Hybrid Search utilizes Reciprocal Rank Fusion algorithms to merge search results.
Hybrid Search is built with BM25
Hybrid Search utilizes the BM25 algorithm for lexical keyword matching.
BigQuery uses Conversational Analytics
BigQuery provides Conversational Analytics agents using functions like AI.AGG
Google developed the Gemma family of open models.
Related events
Google Announces BigQuery Search Innovations for Unstructured Data
Get the weekly recap
The stories like this one, picked and explained — once a week, straight to your inbox.