New Information Retrieval (cs.IR) papers, Tuesday 20 October 2026
41 papers announced on arXiv. Get them ranked for you, free →
When Does Evidence-Graph Structure Predict Answer Correctness? Label-Free Abstention for Structured Retrieval
AI summaryThe paper proposes a training-free abstention method that uses the weighted cut between selected and unselected evidence in a graph to flag unreliable answers. On MemoryAgentBench it improves the precision‑coverage AUC from 0.736 to 0.785 and can identify a 71‑query subset with perfect precision.
Location, Location, Generation: Where You Ask Matters for Generative Engine Optimization
AI summaryGenerative engines tailor cited sources to the user’s declared location rather than network IP. ChatGPT localizes results to the city it is told, while Gemini uses the network country, altering recommended businesses and sources accordingly.
Cache2LoRA: Understanding and Improving Context Internalization
AI summaryCache2LoRA generates low-rank parameter updates from a reusable document cache guided by the query. It matches full-context in-context learning performance across 17 tasks while delivering nearly six times the decoding throughput.
Candidate-dependent residual routing trades top-of-list accuracy for recall in sequential recommendation
AI summaryCandidate‑dependent residual routing adds a lightweight module to a self‑attentive sequential recommender, keeping NDCG@10 unchanged while boosting recall by 1.8% at the expense of a 1.1% drop in MRR, mainly moving hits from positions 1–5 to 6–10.
Optimal Fusion Weights Form Intervals, Not Points: A Latency-Aware Study of Query-Adaptive WRRF
AI summaryQuery‑adaptive fusion weights form intervals, not single points, and most queries exhibit multiple optimal weight ranges. Across eight dynamic strategies, 74 of 128 outperform static RRF; the authors provide a latency‑aware decision framework for selecting appropriate methods.
DuoLM: Fast and Effective Personalized Recommendation with Language Models
AI summaryDuoLM uses two LLMs—one to encode items into compact vectors offline and another to rank candidates using user history. It achieves teacher‑model quality while boosting scoring throughput to 26k candidates/sec, roughly 60× faster than the teacher on an H100.
RAGit: Turning Document Preprocessing into an Experimental Variable for RAG
AI summaryThe authors introduce RAGit, an open-source workbench for systematically evaluating document preprocessing choices in Retrieval-Augmented Generation pipelines. They demonstrate that treating preprocessing as a controlled experimental variable enables reproducible comparisons of parsing and chunking strategies on downstream retrieval performance.
MemRevise: A Self-Revising Memory for Conversational Search
In conversational search, users correct misun-derstandings, revise their requests and return to earlier topics, so part of the history no longer reflects what they want. Existing benchmarks do not isolate such turns. Current search methods either re-read the history or keep a…
VisR-RAG-Fin: Benchmarking Multimodal RAG over Visually Rich Financial Documents
AI summaryVisR‑RAG‑Fin benchmarks multimodal retrieval‑augmented generation on visually rich financial reports, preserving layout and visuals. It shows how text‑only, multimodal, and hybrid retrieval‑generation designs impact QA performance, giving practitioners concrete guidance for enterprise document pipelines.
Learning to Program Retrievals: Composing Search Primitives via Reinforcement Learning
AI summaryRetrieval Program Learning lets a policy compose search primitives into programs with state and control flow, and program mutation generates local variants for better exploration. On QA and corpus reasoning, it outperforms larger baselines, especially on multi‑hop tasks, while reducing context and retrieval steps.
Structure-Aligned History Learning with Subspace-Structured Semantic IDs for Generative Recommendation
AI summaryStructure‑Aligned History Learning (SAHL) uses a balanced token‑subspace‑item hierarchy from its Semantic Residual Balanced Quantization to reconstruct user histories and improve next‑item prediction. SAHL boosts NDCG@10 by 18.6%–30.0% over RPG and 2.7%–6.0% over MHL.
Toward Agentic Retrieval-Augmented Generation for Enterprise Retail Workflows over Heterogeneous Data Sources
AI summaryRetailBridge is a new retail benchmark that simulates complex, heterogeneous data workflows, requiring agents to chain evidence across structured tables and unstructured text. RetailRAG, a framework with schema briefing and source‑routing, outperforms existing baselines and shows multi‑source chaining is the key bottleneck.
RemoteANN: A Native Vector Search Engine for Cloud Object Storage
Vector search engines built on cloud object storage have drawn growing attention from both academia and industry. These systems typically adopt a disaggregated storage-compute architecture, persisting the primary vector payload in object stores (e.g., Amazon S3) while retaining…
When Semantics Fall Short: Persona-Guided Cross-Domain Sequential Recommendation with Large Language Models
Cross-domain sequential recommendation (CDSR) alleviates data sparsity by transferring user behaviors across domains. Recent advances increasingly leverage large language models (LLMs) to bridge domains through global item semantics. However, this paradigm is inherently limited…
Bridging Concepts: Multi-Agent Reasoning over Scientific Concept Graphs for High-Impact Research Prediction
AI summaryThe paper builds scientific concept graphs from conference papers and proposes MiroGraph, a multi‑agent system that designs, extracts, and classifies features to forecast high‑impact concept pairs. MiroGraph outperforms existing impact‑forecasting and temporal link‑prediction baselines.
MEMORA: A Memory-Enhanced Multimodal Committee Reranking Agent for Reasoning-Intensive Retrieval
AI summaryMEMORA splits multimodal queries into intent factors, uses a memory‑conditioned meta‑verdict to rerank, and employs a chain‑of‑thought retriever fine‑tuned for cross‑modal reasoning. On MM‑BRIGHT, it reaches 41.5 nDCG@10 for text‑to‑text and 38.9 for multimodal queries, beating prior baselines by ~8 points.
Where and Why Search-Augmented LLMs Fail When Reasoning over Rapidly Evolving Knowledge
AI summaryTREK is a framework and benchmark that diagnoses why large language models struggle to reason about fast‑changing facts. It shows that the main failure lies in post‑retrieval reasoning, where models get misled by conflicting evidence rather than retrieval errors.
Whistle and Wrong Lyrics: Synthetic Queries Are Not Enough for Real-World Query-by-Melody
People look up a song by humming it, but also by whistling it or singing it with the wrong words. Published query-by-humming systems and benchmarks focus on hums and sung queries, and no public dataset pairs whistles or wrong-lyrics singing with song recordings. We study both…
MemDerive: A Diagnostic Benchmark for Agent Memory over Evolving Data
AI summaryMemDerive is a diagnostic benchmark that records exact derivation DAGs for agent memory over evolving data, enabling error tracing at intermediate states. It shows that existing memory systems lag simple retrieval baselines and rarely propagate implicit changes, highlighting the need for gathering, versioned rules, and dependency maintenance.
Weakly Supervised Adaptation of Scientific Embeddings for Exploratory Literature Mapping
Scientific document embeddings such as SPECTER2 are optimized for retrieval over large, heterogeneous corpora. On a small specialized corpus their similarity distributions can become highly compressed, which limits their usefulness for exploratory literature mapping. We study…
Selling in the Wild: Benchmark and Randomized Field Evidence for a Sales Agent Trained on Commercial Conversations
We test whether Javo, a sales agent fine-tuned on 4.6 million consented, de-identified commercial conversations, sells better than a frontier general-purpose model and better than human sales associates. The evaluation has four parts. Study 1 scores the agent on two public sales…
Benchmark for Structured Retrieval-Augmented Generation in Perovskite Photovoltaics
While retrieval-augmented language models increasingly serve as literature assistants in the physical sciences, existing benchmarks evaluate them primarily on qualitative text overlap or through model-based scoring rather than against verified experimental ground truth. In…
Knowledge Manifold Optimization for Scientific Discovery
The rapid expansion and specialization of scientific literature make it increasingly difficult to identify not only what is known, but where new and still supportable knowledge may emerge. We introduce Knowledge Manifold Optimization (KMO), a discovery-oriented extension of the…
Cross-Pollination Retrieval: Finding Transferable Ideas Across Fields in Scientific Literature
Many scientific advances come from cross-pollination, carrying an idea from one field over to a problem in another. Yet retrieving papers across fields remains underexplored, with existing benchmarks confined to a narrow range of fields. We study cross-pollination retrieval, the…
MAPS: A Multi-Agent Personalized Script-Driven Video Advertising System Enhanced by Reinforcement Learning
With the rapid growth of online video consumption, video advertising has become a dominant component of digital advertising. Most personalized video advertising systems follow a retrieval-based paradigm, which matches ad content with user preferences in a feature space and…
OpenSaC: A Python Library for Search as Code in Deep Research Agents
Deep research agents investigate complex questions through iterative search, source examination, and evidence synthesis. A conventional workflow alternates reasoning with individual search and reading actions. Search as Code introduces a new paradigm: agents write programs that…
Learning Complementary Agent Groups via Task Hypergraphs for Next-POI Recommendation with Missing Modalities
Next point-of-interest (POI) recommendation uses visit histories and POI content, but text and images are often missing. Large language model agents have been applied to recommender systems for preference modeling and evidence acquisition. The challenge is to select agents that…
OneChord: Distill What You Rank, Recommend in One Pass
Accurate product discovery on the Web requires recommendation models that rank catalog candidates within tight inference budgets. In semantic-ID recommendation, deployment can fuse code probabilities with item similarities, making the final ranking depend on their normalization…
Multimodal Decoupling for Diversified Recommendation through Semantic and Structural Shaping
Recommender systems often suffer from over-concentrated results when optimized for accuracy, limiting recommendation diversity. While diversity has become an important objective, existing methods still face fundamental limitations from both semantic and structural perspectives.…
Decision4Rec: A Jev-Inspired Framework of Decision Models on Knowledge and Active Graphs for Multi-Agent Recommender Systems
Decision models, such as the commercial Jev, are language models that return one probability for each option of a question instead of generating text. We present Decision4Rec, an open and modular framework that makes decision models useful for recommendation. A knowledge graph…
Don't Rewrite the SID Map: Topology Preservation for Generative Retrieval in Dynamic E-Commerce Search
Generative retrieval assigns each item a discrete semantic identifier (SID) and retrieves items by autoregressively generating SID sequences. In dynamic environments such as e-commerce, new items arrive continuously, requiring periodic updates to the SID-to-Item mapping and…
FLMNet: Self-Ensemble Distillation from Masks to Seeds for Single-Tower CTR Prediction
Click-through rate (CTR) prediction underpins recommender systems and online advertising, where high-dimensional categorical features interact in combinatorially many ways. Several recent CTR models, such as FinalNet and QNN, report higher accuracy when a network also learns…
Conditional information Retrieval for Longitudinal Survey Harmonisation
Identifying conceptually equivalent questions across different surveys is a challenging information retrieval task where surface-level textual similarity is insufficient. Two questions may be lexically similar yet assess entirely different sociological sub-concepts, or lexically…
Babbar26 main x genre
We introduce Babbar26, a multilingual dataset of 1111 web pages from 1006 domains in five languages (English, French, German, Spanish, Italian), jointly annotated for genre, following a compact five-class taxonomy, and for main content and boilerplate, as main/noise regions…
Preference Is Not Position: Evaluating Reranking on NevIR in Fixed Hybrid Candidate Pools
Near-identical passages can differ in a decisive condition, such as negation. A ranker may order such a pair correctly while placing the preferred passage far down the retrieved list. We examine this distinction using official NevIR pairs jointly covered by fixed candidate pools…
The Scheduler Is a Hidden Retriever: How Asynchrony Shapes Evidence Acquisition in Search Agents
Language-model agents increasingly solve complex tasks through parallel reasoning and asynchronous information gathering. Controlled replay across two model families and three benchmarks shows that delivery timing redirects search decisions made before all results arrive,…
Large-Scale Biomedical RAG: Corpus Scaling, Document Representation, and Efficiency Trade-offs
We present a controlled, large-scale evaluation of how corpus scale, document representation, and reranking affect the effectiveness and efficiency of biomedical Retrieval-Augmented Generation (RAG) systems. Retrieval collections were progressively expanded while retaining the…
The Geometry of Agent Memory: Forecasting Retrieval-Bound Memory Failures, and Characterizing Its Limits
AI summaryThe paper builds a shared embedding space for conversation turns, memories, and queries and trains a logistic classifier to predict answer correctness from 25 geometric features. It achieves an AUROC of 0.640 (vs. 0.655 baseline), adding negligible improvement beyond question difficulty and system identity.
Beyond Behavioral Accuracy: Evaluating Individual-Level Fidelity in LLM Investor Simulation
Large language models (LLMs) are increasingly used as user simulators, but can they faithfully reproduce an individual's behavior and decision logic over time? We investigate this question in a longitudinal paper-trading study with 80 participants. Using rolling next-day…
Next Best Action: Causal Constrained Offline RL for Business Action Targeting
Next Best Action (NBA) systems must select among customer interventions (e.g., discounts, promotions, and lifecycle actions), maximize long-term economic value, satisfy complex operational constraints, and remain causally grounded. Existing approaches lack a rigorous framework…
AGR: Agentic Generative Retrieval with Tool-Augmented Search
Large-scale search and recommendation systems derive much of their retrieval quality from rich, heterogeneous features, yet existing generative retrieval (GR) methods struggle to exploit such feature ecosystems efficiently. Static feature injection either encodes all available…