This repository is a curated list of resources for the field of generative information retrieval, focusing on conversational AI that can access external information and cite sources. It serves researchers and practitioners interested in advanced AI systems that function as information retrieval machines, offering a comprehensive overview of papers, datasets, tools, and workshops.

How It Works

The field is broadly categorized into "Grounded Answer Generation" and "Generative Document Retrieval." Grounded Answer Generation encompasses techniques like Retrieval Augmented Generation (RAG), LLM memory manipulation, and self-correction for factual accuracy. Generative Document Retrieval focuses on methods for generating document identifiers or strings that facilitate efficient retrieval, including approaches like differentiable search indexes and autoregressive models.

Quick Start & Requirements

This repository is a collection of links and does not have a direct installation or execution command. Users will need to follow the links provided for individual papers, datasets, and tools to access their respective requirements and setup instructions.

Highlighted Details

• Extensive categorization of research topics within generative information retrieval.
• Comprehensive lists of relevant academic papers, datasets, and tools with links.
• Includes a "Live Generative Retrieval" section showcasing active demos and services.
• Covers emerging areas like generative recommendation and knowledge graphs.

Maintenance & Community

The repository is community-driven with a call for pull requests to add new resources. Specific contributors or maintainers are not highlighted, but the extensive list of cited works suggests active community engagement.

Licensing & Compatibility

The repository itself is a list of links and does not have a specific license. The licensing of individual papers, datasets, and tools will vary and must be checked on their respective sources.

Limitations & Caveats

As a curated list, the repository's content is dependent on the community's contributions and may not be exhaustive or perfectly up-to-date. The rapid evolution of the field means some linked resources might become outdated.