Introduction to Google’s New Search Methods
Google recently announced two new methods to improve search results: Multi-Vector Retrieval via Fixed-Dimensional Encodings (MUVERA) and Graph Foundation Models (GFMs). These methods aim to improve the way Google retrieves and ranks search results. In a recent Search Central Live Deep Dive in Asia, Google’s Gary Illyes answered questions about whether these methods are being used in Google’s search algorithm.
What is MUVERA?
MUVERA is a method that improves retrieval by turning complex multi-vector search into fast single-vector search. It compresses sets of token embeddings into fixed-dimensional vectors that closely approximate their original similarity. This allows Google to use optimized single-vector search methods to quickly find good candidates, then re-rank them using exact multi-vector similarity. Compared to older systems, MUVERA is faster, retrieves fewer candidates, and still improves recall, making it a practical solution for large-scale retrieval.
Key Points About MUVERA
- MUVERA converts multi-vector sets into fixed vectors using Fixed Dimensional Encodings (FDEs), which are single-vector representations of multi-vector sets.
- These FDEs match the original multi-vector comparisons closely enough to support accurate retrieval.
- MUVERA retrieval uses MIPS (Maximum Inner Product Search), an established search technique used in retrieval, making it easier to deploy at scale.
- Reranking: After using fast single-vector search (MIPS) to quickly narrow down the most likely matches, MUVERA re-ranks them using Chamfer similarity, a more detailed multi-vector comparison method.
- MUVERA is able to find more of the precisely relevant documents with a lower processing time than the state-of-the-art retrieval baseline it was compared to.
Google Confirms Use of MUVERA
Gary Illyes confirmed that Google uses a version of MUVERA in their search algorithm. When asked about MUVERA, Illyes jokingly asked what it was, then confirmed that they use something similar to MUVERA, but don’t name it as such.
What are Graph Foundation Models?
Graph Foundation Models (GFMs) are a type of AI that learns from relational databases by turning them into graphs, where rows become nodes and the connections between tables become edges. Unlike older models, GFMs can handle new databases with different structures and features without retraining on the new data. GFMs use a large AI model to learn how data points relate across tables, allowing them to find patterns that regular models miss.
Key Points About GFMs
- GFMs can handle new databases with different structures and features without retraining on the new data.
- GFMs use a large AI model to learn how data points relate across tables.
- GFMs can find patterns that regular models miss, and perform much better in tasks like detecting spam.
- GFMs represent a notable achievement, with performance gains of 3x to 40x in average precision.
Is GFM Ready for Scaled Deployment?
The official Graph Foundation Model announcement says it was tested in an internal task, spam detection in ads, which strongly suggests that real internal systems and data were used, not just academic benchmarks or simulations. However, Gary Illyes expressed his opinion that GFMs are not currently used in Google’s search algorithm.
Takeaways
Google’s Gary Illyes confirmed that a form of MUVERA is in use at Google. His answer about GFMs seemed to be expressed as an opinion, so it’s somewhat less clear. While we can’t say for certain whether GFMs are being used in Google’s search algorithm, it’s clear that Google is continually working to improve their search results using new and innovative methods.
Conclusion
In conclusion, Google’s new search methods, MUVERA and GFMs, have the potential to greatly improve search results. While we know that MUVERA is being used, the use of GFMs is less clear. As Google continues to develop and refine these methods, we can expect to see even more accurate and relevant search results in the future.