Information Retrieval
Papers on search engines, recommendation systems, and information extraction.
cs.IR · 379 papersExpressiveness Limits of Autoregressive Semantic ID Generation in Generative Recommendation
Generative recommendation models' autoregressive ID generation limits expressiveness due to tree-structured decoding, which Latte mitigates for better performance.
Addressing Labelled Data Scarcity: Taxonomy-Agnostic Annotation of PII Values in HTTP Traffic using LLMs
This paper introduces an LLM-based pipeline for taxonomy-agnostic PII annotation in HTTP traffic, addressing data scarcity and evolving privacy definitions.
OBLIQ-Bench: Exposing Overlooked Bottlenecks in Modern Retrievers with Latent and Implicit Queries
OBLIQ-Bench introduces a new benchmark for "oblique" queries, revealing that modern retrievers struggle to find documents with latent patterns, unlike LLMs.
Revisiting Uncertainty: On Evidential Learning for Partially Relevant Video Retrieval
Holmes introduces a hierarchical evidential learning framework to explicitly model and quantify uncertainty in partially relevant video retrieval, outperforming SOTA.
A Case-Driven Multi-Agent Framework for E-Commerce Search Relevance
A multi-agent framework automates e-commerce search relevance optimization by replacing human roles with specialized AI agents for case identification and resolution.
Bridging Passive and Active: Enhancing Conversation Starter Recommendation via Active Expression Modeling
PA-Bridge enhances conversation starter recommendations by using active user expressions and an adversarial aligner to overcome feedback loop issues.
Unified Value Alignment for Generative Recommendation in Industrial Advertising
UniVA enhances generative recommendation for advertising by aligning commercial value signals across tokenization, decoding, and online serving.
Beyond Long Tail POIs: Transition-Centered Generalization for Human Mobility Prediction
RECAP improves human mobility prediction by addressing transition-level sparsity, reconstructing rare POI transitions for better generalization.
Effective Knowledge Transfer for Multi-Task Recommendation Models
EKTM improves multi-task recommendation by transferring knowledge across CVR tasks, boosting conversion rates and platform effectiveness.
Text-Graph Synergy: A Bidirectional Verification and Completion Framework for RAG
TGS-RAG is a novel framework that uses bidirectional text-graph synergy to improve RAG by refining textual evidence and resurrecting pruned graph paths.
AgenticRAG: Agentic Retrieval for Enterprise Knowledge Bases
AgenticRAG improves enterprise RAG by using an LLM agent with tools for iterative retrieval and analysis, significantly boosting recall and factuality.
Open-SAT: LLM-Guided Query Embedding Refinement for Open-Vocabulary Object Retrieval in Satellite Imagery
Open-SAT improves open-vocabulary satellite image retrieval by using LLMs to refine query embeddings at inference time, achieving significant F1 score gains.
Securing the Agent: Vendor-Neutral, Multitenant Enterprise Retrieval and Tool Use
This paper introduces a layered isolation architecture to secure multitenant enterprise RAG and agentic AI systems, preventing data leakage.
Interests Burn-down Diffusion Process for Personalized Collaborative Filtering
A new "interests burn-down diffusion process" is proposed for collaborative filtering, better modeling user interest decay for recommendations.
CapsID: Soft-Routed Variable-Length Semantic IDs for Generative Recommendation
CapsID introduces soft-routed, variable-length Semantic IDs for generative recommendation, significantly improving recall and efficiency over existing methods.
Empirical Study of Pop and Jazz Mix Ratios for Genre-Adaptive Chord Generation
This paper investigates optimal data mix ratios for fine-tuning a pop-trained chord generation model to jazz, balancing new genre acquisition with old genre retention.
TabEmbed: Benchmarking and Learning Generalist Embeddings for Tabular Understanding
TabEmbed introduces a generalist embedding model for tabular data, unifying classification and retrieval, alongside TabBench for evaluation.
Storage Is Not Memory: A Retrieval-Centered Architecture for Agent Recall
True Memory introduces a retrieval-centered architecture for agent recall, achieving high accuracy by preserving verbatim events and outperforming existing systems.
RecGPT-Mobile: On-Device Large Language Models for User Intent Understanding in Taobao Feed Recommendation
RecGPT-Mobile deploys lightweight LLMs directly on mobile devices to understand user intent in real-time, improving e-commerce recommendations.
Rethinking Convolutional Networks for Attribute-Aware Sequential Recommendation
ConvRec introduces a convolution-based model for attribute-aware sequential recommendation, achieving efficiency and outperforming attention methods.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.