Searchable index
This Google patent describes a system and method for creating a searchable index based on machine learning models. The technology generates index entries containing tokens correlated with outcomes and their read more
Crawl algorithm
This Google patent describes a web crawler algorithm that efficiently manages and updates cached web pages while optimizing resource usage. The system determines crawl values for web pages based on read more
LLM Alignment as Retriever Optimization: An Information Retrieval Perspective
This Google Deepmind paper presents a novel approach to Large Language Model (LLM) alignment by drawing parallels with Information Retrieval (IR) systems. The authors introduce LarPO (LLM Alignment as Retriever read more
Generating Categories for Sets of Entities
This research paper presents a system that can automatically generate new, fine-grained categories for sets of related entities—such as items in a Wikipedia table—especially when no suitable category currently exists. read more
Rankers, Judges, and Assistants: Towards Understanding the Interplay of LLMs in Information Retrieval Evaluation
This Google Deepmind research paper examines the complex interplay between different uses of Large Language Models in information retrieval systems, particularly focusing on their roles as rankers, judges, and content read more
GINGER: Grounded Information Nugget-Based Generation of Responses
This paper introduces GINGER (Grounded Information Nugget-Based Generation of Responses), a novel approach for generating accurate and verifiable responses in retrieval-augmented generation systems. The system works by breaking down retrieved read more
Selecting an algorithm for identifying similar user identifiers based on predicted click-through-rate
This Google patent describes a computerized method for selecting the most effective algorithm to identify similar user identifiers based on predicted click-through rates. The system compares different algorithms by generating read more
Sufficient Context: A New Lens on Retrieval Augmented Generation Systems
This Google research paper examines the performance of various large language models (LLMs) in question-answering tasks, particularly focusing on their ability to handle sufficient and insufficient context scenarios. The study read more
Machine learning models as a differentiable search index for directly predicting resource retrieval results
This Google patent describes a novel approach to information retrieval using machine learning models called Differentiable Search Index (DSI). The system directly predicts relevant resources in response to queries by read more
Using user input to adapt search results provided for presentation to the user
This Google patent describes a system and method for adapting search results presented by an automated assistant during a user dialog. The technology enables users to navigate through search results read more