Information Retrieval
Papers on search engines, recommendation systems, and information extraction.
cs.IR · 379 papersBeyond Static Best-of-N: Bayesian List-wise Alignment for LLM-based Recommendation
BLADE introduces a Bayesian framework for LLM-based recommenders to dynamically optimize list-wise metrics, outperforming static methods.
DoGMaTiQ: Automated Generation of Question-and-Answer Nuggets for Report Evaluation
DoGMaTiQ automates the generation of high-quality, QA-based "nuggets" for evaluating RAG reports, showing strong correlation with human judgments.
One Pool, Two Caches: Adaptive HBM Partitioning for Accelerating Generative Recommender Serving
HELM adaptively partitions GPU HBM between embedding and KV caches for generative recommenders, reducing P99 latency by 24-38% across diverse workloads.
Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems
This paper introduces BRIGHT-Pro, a new benchmark, and RTriever-Synth, a training corpus, to advance reasoning-intensive retrieval for agentic search systems.
Domain-Adaptive Dense Retrieval for Brazilian Legal Search
This paper explores domain-adaptive dense retrieval for Brazilian legal search, finding a mixed training approach offers robust performance.
Physics-Grounded Multi-Agent Architecture for Traceable, Risk-Aware Human-AI Decision Support in Manufacturing
MAKA is a physics-grounded multi-agent AI architecture for traceable, risk-aware human-AI decision support in high-precision manufacturing.
Aspect-Aware Content-Based Recommendations for Mathematical Research Papers
This paper introduces AchGNN, an aspect-conditioned GNN, and new datasets for content-based mathematical research paper recommendations, outperforming prior methods.
Cosmodoit: A Python Package for Adaptive, Efficient Pipelining of Feature Extraction from Performed Music
Cosmodoit is a Python package that streamlines feature extraction from performed music by integrating various algorithms into an efficient, modular pipeline.
SURE-RAG: Sufficiency and Uncertainty-Aware Evidence Verification for Selective Retrieval-Augmented Generation
SURE-RAG improves Retrieval-Augmented Generation by verifying evidence sufficiency, reducing unsafe answers through a transparent aggregation protocol.
Revisiting General Map Search via Generative Point-of-Interest Retrieval
GenPOI is a generative framework using LLMs to improve map search by handling underspecified queries through spatial-aware POI retrieval.
RAG over Thinking Traces Can Improve Reasoning Tasks
This paper shows that using "thinking traces" as a retrieval corpus significantly enhances RAG performance on complex reasoning tasks like math and code.
Beyond Similarity Search: A Unified Data Layer for Production RAG Systems
This paper proposes a unified PostgreSQL-based data layer for RAG systems, significantly improving reliability, performance, and security.
AlbumFill: Album-Guided Reasoning and Retrieval for Personalized Image Completion
AlbumFill is a training-free framework that retrieves identity-consistent references from personal albums for personalized image completion.
Multi-Axis Speech Similarity via Factor-Partitioned Embeddings
This paper introduces factor-partitioned embeddings to disentangle speech attributes like content and speaker identity, enabling multi-axis similarity for improved retrieval.
Benchmarking Retrieval Strategies for Biomedical Retrieval-Augmented Generation: A Controlled Empirical Study
This paper systematically compares five retrieval strategies for biomedical RAG, finding Cross-Encoder Reranking performs best.
From Experimental Limits to Physical Insight: A Retrieval-Augmented Multi-Agent Framework for Interpreting Searches Beyond the Standard Model
HEP-CoPilot is a retrieval-augmented multi-agent AI framework that unifies diverse high-energy physics data to accelerate BSM search interpretation.
GRAIL: A Deep-Granularity Hybrid Resonance Framework for Real-Time Agent Discovery via SLM-Enhanced Indexing
GRAIL is a novel framework for real-time agent discovery, achieving sub-400ms latency and high accuracy using SLMs and fine-grained matching.
FitText: Evolving Agent Tool Ecologies via Memetic Retrieval
FitText introduces a training-free framework that dynamically evolves agent tool retrieval by generating and refining pseudo-tool descriptions using memetic retrieval.
Is It Novel and Why? Fine-Grained Patent Novelty Prediction Based on Passage Retrieval
This paper introduces FiNE-Patents, a new dataset and LLM-based approach for fine-grained patent novelty prediction via passage retrieval.
Fair Agents: Balancing Multistakeholder Alignment in Multi-Agent Personalization Systems
This paper proposes a conceptual framework for designing fair multi-agent personalization systems that balance competing stakeholder objectives using LLM agents.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.