arXiv:2605.00505v1 [cs.IR] 1 May 2026
LLM-Oriented Information Retrieval: A Denoising-First Perspective Lu Dai
Liang Sun
Fanpu Cao
[email protected] Hong Kong University of Science and Technology Hong Kong SAR, Hong Kong
Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
Ziyang Rao
Cehao Yang
Hao Liu
Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
[email protected] Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China
Hui Xiong [email protected] Hong Kong University of Science and Technology (Guangzhou) Guangzhou, China Hong Kong University of Science and Technology Hong Kong SAR, Hong Kong
Abstract
Keywords
Modern information retrieval (IR) is no longer consumed primarily by humans but increasingly by large language models (LLMs) via retrieval-augmented generation (RAG) and agentic search. Unlike human users, LLMs are constrained by limited attention budgets and are uniquely vulnerable to noise; misleading or irrelevant information is no longer just a nuisance, but a direct cause of hallucinations and reasoning failures. In this perspective paper, we argue that denoising—maximizing usable evidence density and verifiability within a context window—is becoming the primary bottleneck across the full information access pipeline. We conceptualize this paradigm shift through a four-stage framework of IR challenges: from inaccessible to undiscoverable, to misaligned, and finally to unverifiable. Furthermore, we provide a pipeline-organized taxonomy of signal-to-noise optimization techniques, spanning indexing, retrieval, context engineering, verification, and agentic workflow. We also present research works on information denoising in domains that rely heavily on retrieval such as lifelong assistant, coding agent, deep research, and multimodal understanding.
LLM-oriented information retrieval, information denoising, retrievalaugmented generation, agentic retrieval, faithfulness verification, hallucination mitigation
CCS Concepts • Information systems → Retrieval tasks and goals; Retrieval models and ranking; • Computing methodologies → Natural language processing.
This work is licensed under a Creative Commons Attribution 4.0 International License. SIGIR ’26, Melbourne, VIC, Australia © 2026 Copyright held by the owner/author(s). ACM ISBN 979-8-4007-2599-9/2026/07 https://doi.org/10.1145/3805712.3808544
ACM Reference Format: Lu Dai, Liang Sun, Fanpu Cao, Ziyang Rao, Cehao Yang, Hao Liu, and Hui Xiong. 2026. LLM-Oriented Information Retrieval: A Denoising-First Perspective. In Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’26), July 20– 24, 2026, Melbourne, VIC, Australia. ACM, New York, NY, USA, 16 pages. https://doi.org/10.1145/3805712.3808544
1
Introduction
Large-language models (LLMs) have emerged as the new consumers of retrieved information and transformed information access. Rather than manually navigating web pages, users increasingly delegate their search, summarization and reasoning tasks to LLM agents [8, 202]. Accordingly, retrieval has become a crucial interleaving step in an LLM-centric task-solving pipeline, supplying the model with external knowledge to use in reasoning or generation. This paradigm – often referred to as retrieval-augmented generation (RAG) – tightly couples search with generation to improve factual coverage and answer quality [86]. As LLMs take on this role of information consumer, the objectives of information retrieval shift. Classic IR research prioritized metrics like recall and precision – focusing on minimizing missed relevant documents and down-ranking irrelevant ones. In LLM-mediated systems, however, they are necessary but no longer sufficient. One key bottleneck is verifiability. If the search component fails to find the correct supporting information or retrieves misleading
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
Lu Dai et al.
Consumer of Retrieved Informatinon: Human à LLM
Inverted Index
Library
Card Catalog
PageRank
Crawling
CLIP
Bi-Encoder
Unverifiable
Misaligned
Undiscoverable Inaccessible
Project Code
Neural Sparse
Generative Retrieval
Web Search
Lifelong Memory Multimodal Information
Constraint: Physical Reachability
Constraint: Web-Scale Latency
Constraint: Semantic Understanding
Constraint: Attention & Verification
Era 1: Pre-Internet
Era 2: Web Scale
Era 3: Neural IR
Era 4: LLM-Oriented IR
Figure 1: Challenge shifts in the history of IR. information, even a powerful LLM cannot produce a correct and verifiable answer. On the one hand, LLM-generated content is flooding the internet corpus itself. The proliferation of hallucinations makes attribution and trust harder than ever before. On the other hand, LLMs are sensitive to noise in context. Studies have found that misleading evidence in the context can be more detrimental to an LLM’s answer quality than missing evidence [3]. These findings highlight that modern retrieval must focus not just on finding more information, but on finding reliable information that the LLM can faithfully use and verify as evidence. Another critical challenge is the limited attention and context capacity of current LLMs. Even with the latest models featuring context windows extended to millions of tokens, LLMs struggle to fully utilize very large contexts, especially in reasoning tasks [15, 100]. Empirical results confirm that beyond a certain point, adding more retrieved passages, especially irrelevant and redundant ones, yields diminishing or even negative returns. Furthermore, noisy or malicious content in the context can actively undermine the model’s reliability [28]. A well-known vulnerability is prompt injection, where adversarially crafted text in the retrieved context hijacks the model’s behavior or injects false information [102]. This highlights that irrelevant or harmful information in the context window isn’t just a waste of computation resources – it can directly degrade output quality or even pose serious risks. These observations suggest that the key challenge of modern retrieval is not to retrieve more but to denoise—to provide concise, high-quality context that fits within the model’s attention budget. An LLM-based assistant might call a retriever frequently and aggregate many evidence snippets in one session; even a small fraction of noise can cascade through such a pipeline, leading to grounded hallucinations or degraded performance. While LLMs do exhibit some robustness, it is risky to rely on the model alone to filter out
noise. Instead, the retrieval stage itself should take on the burden of ensuring a high signal-to-noise ratio in the provided context. In this paper, we present the perspective of denoising as the primary challenge for information retrieval in the era of LLM. We first characterize the challenge shift through the history of information retrieval into four stages and validate the importance of denoising with experimental evidence. We then present a comprehensive taxonomy of research efforts on denoising across the whole retrieval-augmented generation pipeline, as well as curating solutions in typical information-intensive and long-context application areas. Finally, we summarize the solutions and limitations of current LLM-oriented information retrieval and propose potential future directions. Through this paper, we hope to shed light on the challenge shift of information retrieval in LLM era towards an emphasis on utility and verifiability, and inspire future innovations in this important research direction.
2
Challenge shifts in the history of IR
We conceptualize the evolution of IR as a progression of dominant bottlenecks—specific constraints that made information retrieval fail in distinct ways. As illustrated in Figure 1, the field has transitioned through four phases: from information inaccessible to undiscoverable, to misaligned, and finally to unverifiable. Era 1 (Pre-internet): Inaccessible under physical constraints. Before the World Wide Web, the “evidence universe” was inherently local. Information was not primarily unranked, but unreachable—constrained by geography, institutional access, and physical latency. In this era, the dominant bottleneck was availability rather than discoverability; the challenge was not selecting the best document, but physically acquiring any relevant document at all [12]. Era 2 (Web-scale IR): Undiscoverable under indexing scale. The web inverted the bottleneck: information became globally
LLM-Oriented Information Retrieval: A Denoising-First Perspective
reachable but overwhelming. The challenge shifted to undiscoverability—valid evidence existed but was buried under an evergrowing, unindexed corpus. In this stage, the core objective became efficient crawling, indexing, and ranking to surface signals from noise at interactive latencies. Apart from relevance, quality estimation emerged as a critical filter, with PageRank leveraging hyperlink structures as a scalable proxy for authority to mitigate web noise [9]. Importantly, this era institutionalized large-scale evaluation infrastructure. Benchmarks and shared tasks (e.g., TREC) normalized reproducible experiments on large collections and anchored the field around measurable improvements under web-scale constraints. Lexical matching (e.g., BM25) became the industrial standard, establishing a robust effectiveness–efficiency frontier that remains relevant today [143]. Era 3 (Neural IR): Misaligned under semantic constraints. Once IR systems could reliably retrieve candidates at scale, the semantic barrier became the primary problem: even when evidence is retrieved, systems may not understand, compose, or reason with it. Besides, lexical overlaps often failed to capture user intent, leaving retrieved documents misaligned with semantic needs. Research works in this era focused on bridging the semantic gap [164]. Dense retrieval reduced the lexical mismatch between queries and relevant documents by learning continuous semantic representations (e.g., DPR) [77]. Neural rankers improved relevance modeling but introduced steep inference costs; late-interaction architectures such as ColBERT offered a structural compromise between cross-encoder fidelity and bi-encoder efficiency [79]. Meanwhile, neural sparse methods (e.g., SPLADE) reframed lexical matching itself as a learnable expansion mechanism, blurring the sparse-dense dichotomy [39]. Generative models also entered relevance modeling: MonoT5 cast reranking as a seq2seq prediction problem, using generative likelihood as a relevance signal [124]. In this era, IR systems began to deeply integrate with other NLP tasks under the retrieval-augmented generation scheme, setting the stage for end-to-end problem solving [67, 86]. Era 4 (LLM-Oriented IR): Unverifiable context overload. We represent the current era not merely as an application shift, but as a fundamental redefinition of the retrieval objective driven by two simultaneous transformations. First, LLMs have turned retrieval from a destination into an input channel. In retrieval-augmented generation, the consumer is no longer a human browsing for links, but a model requiring actionable evidence to drive generation [86, 174]. Retrieval success is thus redefined by downstream utility: a document is only useful if it positively influences the model’s reasoning trace [28, 145]. Second, LLM-generated content is flooding the corpus. As the web becomes saturated with synthetic text, retrieval faces a crisis of epistemic integrity. The recursive training on generated content threatens model collapse [155], creating a homogeneous and potentially adversarial search space where source bias and noise actively compete with factual signals [29]. These shifts expose a fundamental friction: while traditional IR strives for high recall, LLMs struggle to filter the resulting noise from their limited attention span. We identify three intrinsic vulnerabilities where retrieval artifacts directly degrade reasoning: Observation 1: Fragmentation brings in conflicts. Retrieving and assembling snippets from disparate sources often strips away their
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
original discourse context, such as temporal validity or conditional scope. This de-contextualization creates a "Frankenstein" context where conflicting or outdated information appears semantically relevant. Such semantic noise actively competes with correct evidence for attention, confusing the model’s selection mechanism even when the context length is manageable [169, 195]. Observation 2: Context Dilution. Extended context windows do not guarantee utilization. As the retrieved context grows, useful information becomes buried or "lost-in-the-middle," exhausting the model’s attention budget. This dilution effect often leads to reasoning failures even when the answer is present in the context [7, 100]. Observation 3: Cascading Failures. With the rise of agentic and multistep reasoning workflows, the cost of retrieval errors has escalated. A single misleading or conflicting snippet can propagate through the reasoning chain, converting retrieval noise into compounded hallucinations that undermine the entire task execution [122]. Consequently, we argue that the primary bottleneck in Era 4 has shifted from access to utility. To prevent these upstream noises from disrupting downstream reasoning, the retrieval system must evolve into a noise gate, prioritizing the maximization of signal-to-noise ratio and verifiability over raw recall.
2.1
Empirical Validation: Noise as Bottleneck
We validated the "denoising-first" perspective using LLaMA-2-7BChat on 500 Natural Questions (NQ) samples, each paired with 100 DPR-retrieved passages (labeled as gold or noise). Impact of Context SNR. We systematically vary the signal-tonoise ratio of retrieved context. As shown in Figure 2(a), EM improves monotonically from 47.0% to 61.0% when increasing gold passages from 1 to 5. However, noise erodes this gain rapidly: holding gold passages fixed at 3, adding 2 and 7 noise passages reduces EM to 51.4% and 41.8%, respectively. When a single gold passage is buried among 9 noise passages (SNR=0.10), EM falls to 26.6%, barely above the closed-book baseline of 23.6%. Pure noise contexts yield only 8.0% EM—far worse than no retrieval—demonstrating that noise actively impairs the model’s parametric memory. Positional Bias vs. Noise. Fixing the context to 1 gold and 9 noise passages while varying the gold passage’s position (Figure 2(b)), we find that noise-induced degradation dominates positional effects. A single gold passage alone achieves 43.2% EM, but adding 9 noise
Figure 2: Empirical validation of the denoising-first perspective on NQ. (a) SNR–utility curve: as the gold passage ratio decreases, EM degrades sharply. (b) Positional sensitivity: the ∼20% drop from noise >> the ∼4% positional variation.
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
passages collapses performance to 22.2–26.6% regardless of position. A mild U-shaped positional pattern is visible but dwarfed by the noise effect.
3
Denoising in LLM-Oriented Retrieval: A Method Taxonomy
In the context of LLM-oriented IR, the retrieval objective fundamentally shifts from prioritizing recall to acting as a noise gate. We use noise to refer to any retrieved content that consumes budget (tokens, latency, attention) without helping—or that actively misleads the model. Drawing on the shift described in Section 2, we categorize the primary entry points of noise in LLM-oriented IR into three distinct classes: Corpus-level noise. Noise originates upstream within the knowledge source itself, where indices are increasingly contaminated by duplicates, template spam, or outdated information [94]. Recent studies highlight a growing risk of “spiral” feedback loops, where AI-generated content becomes overrepresented in retrieval pools, potentially destabilizing provenance [19]. Furthermore, even minor injections of adversarial misinformation or low-credibility evidence can severely degrade open-domain QA, as standard retrievers lack the mechanisms to verify truthfulness before indexing [129]. Retriever-level noise. Even with a clean corpus, retrievers introduce noise by surfacing “hard distractors”—passages that are topically relevant but factually non-supporting or misleading [195]. Dense retrievers are particularly prone to these semantic “near misses,” which occupy the embedding space near the query but fail to answer it [3, 27]. Research indicates that these non-informative contexts actively divert the model’s attention, leading to distractioninduced hallucinations and reduced accuracy even when correct evidence is present [100]. Context construction noise. Noise is often amplified in the last mile between retrieval and generation, when candidates are chunked, merged, and ordered into a single prompt. First, structural and positional noise stems from the mismatch between retrieval ranking and LLM attention; naive concatenation often buries critical evidence in the “lost-in-the-middle” zone, rendering it effectively invisible to the model [100, 178]. Second, evidence incoherence arises when the retriever aggregates conflicting or redundant passages; this incoherence increases parametric uncertainty and triggers hallucinations, requiring diversity-aware compression rather than simple concatenation [170, 187]. Third, the retrieval channel introduces adversarial noise via indirect prompt injections, where malicious instructions embedded in untrusted documents hijack the generation logic, breaching the system’s safety alignment [49]. Consequently, the IR system must treat evidence assembly as an active denoising stage—responsible for curating, ordering, and sanitizing the prompt—rather than a passive data hand-off. To systematically address these failure modes, we organize our taxonomy of denoising methods aligning with the lifecycle of information flow. The following sections review interventions at the level of Representation (purifying the index), Search (filtering via precision ranking), Context Assembly (optimizing input composition), and Verification (auditing the generative output). Finally, we examine Agentic Loops, where feedback dynamically refines
Lu Dai et al.
the signal through iterative retrieval and self-correction. Figure 3 summarizes this taxonomy in a stage-aligned pipeline view.
3.1
Controlled Indexing
This subsection surveys controlled indexing—index-time interventions that reduce noise by (i) regulating what enters the retrievable universe and (ii) attaching conditionable representations and metadata (e.g., trust, time, structure, safety) that downstream retrieval and context assembly can reliably exploit. Governed indexing sets the retrieval quality ceiling, as downstream filtering cannot fully eliminate a polluted index’s residual noise and overhead. (1) Provenance and Trust Stratification. Rather than treating the corpus as a monolithic pool, effective denoising begins by stratifying the index based on verifyable provenance. This involves capturing granular metadata—such as publisher authority, extraction timestamps, and cryptographic signatures (e.g., C2PA standards [24])—to enable trust-conditioned retrieval, where the search space can be dynamically restricted to high-credibility strata. A critical challenge in the LLM era is distinguishing human insights from machine-generated content to prevent degradation caused by widespread AI content on the open web [19, 29, 94, 155]. To this end, indexing pipelines must integrate synthetic attribution mechanisms: strictly cataloging proactive watermarks [30, 81]; embedding posthoc detection scores [46, 92, 117] as filterable metadata; and flagging copyrighted or otherwise high-risk content for downstream gating [206, 207]. By operationalizing provenance as an admission criterion rather than an afterthought, the system prevents lowcredibility signals from ever competing for the retriever’s attention. (2) Quality Filtering and Canonicalization. Data filtering is not only important for LLM pretraining, but also for retrieval index building, where even small fractions of irrelevant or low-quality passages have been shown to substantially degrade RAG generation quality [26]. To maximize the information density of the index, this family of methods eliminates semantic redundancy and low-quality artifacts that otherwise act as distractors in the embedding space. Deduplication is one of the levers: surface-form resemblance measures (e.g., MinHash) prune near-duplicates [10, 84], while embedding-based semantic deduplication (e.g., SemDeDup) further collapses paraphrastic clusters that lexical hashing misses [1], ensuring the index stores only canonical prototypes rather than repetitive template variations and preventing budget waste during context assembly. Furthermore, data sanitation pipelines established for LLM pre-training provide transferable primitives for index purification. These techniques span language identification, heuristic filtering, and safety scoring as seen in works like RefinedWeb [88, 132, 133, 157, 183]. Adapting these protocols ensures that the retrieval pool is chemically pure, reducing the probability of hallucination-inducing nonsense. (3) Temporal Validity Management. In dynamic environments, semantic relevance no longer implies factual currency, motivating indexing pipelines that treat temporal validity as a first-class admission constraint rather than a static snapshot [127]. Recent designs realize this through time-aware graph structures: VersionRAG and EraRAG maintain explicit version graphs with incremental updates
LLM-Oriented Information Retrieval: A Denoising-First Perspective
Controlled Indexing (§3.1)
Provenance & Trust
Watermark [81]; SynthID [30]; DetectGPT [117]; MAGE [92]; Spiral [19]; Cocktail [29]; GLTR [46]; Model collapse [155]; LLM adoption [94]; CopyrightDetective [206, 207]
Quality Canonicalization
MinHash [10]; Dedup data [84]; CCNet [183]; RefinedWeb [133]; FineWeb [132]; Dolma [157]
Temporal Validity
HOH [127]; VersionRAG [64]; EraRAG [205]; Theanine [125]; STAR-RAG [223]; T-GRAG [87]
Structure-aware Indexing
GraphRAG [36]; GRAG [59]; LightRAG [52]; KG2RAG [222]; G-Retriever [57]; EDC [203]; RAKG [209]; RAPTOR [148]; PageIndex [211]; HippoRAG [53]
Query Transformation
HyDE [43]; Query2Doc [173]; Step-Back [216]; GenRead [198]; Self-Ask [137]; Enhance-QR [194]; QR-RAG [106]; RaFe [111]; DMQR-RAG [93]; Converser [60]; Decomposition [110]; MMLF [82]; RQRAG [14]; RL rewrite [72, 134]
Hybrid Retrieval
Hybrid retrieval [11]; SPLADE [39]; BGE-M3 [16]; OOD Hybrid [18]; Adaptive routing [109]; Unsupervised Dense IR [66]
Distractor-aware Training
ANCE [188]; RocketQA [140]; SimLM [172]; AR2 [208]; DRAGON [96]; INSTRUCTOR [159]; E5 [171]; RA-DIT [97]; Inst. tuning [27, 177]; RbFT [167]
Fine-grained Matching
ColBERT [79]; ColBERTv2 [146]; XTR [83]; MUVERA [32]; ColPali [38]
Selection & Pruning
Pairwise/Setwise/Listwise reranking [138, 142, 224]; MMR [13]; DPP [147]; Speculative selection [179]; FILCO [176]; Provence [23]; DSLR [65]; Adaptive-RAG [69]
Context Compression
LLMLingua [70]; LongLLMLingua [71]; LLMLingua-2 [130]; RECOMP [189]; Compression for RAG [91]; Gist Tokens [119]; ICAE [45]; xRAG [22]; Token-efficiency objective [190]
Layout Optimization
Lost in the Middle [100]; Topology-aware ordering [136]; OP-RAG [197]; Irrelevant-context robustness [151]
Conflict Resolution
Conflict-aware assembly [170, 187]; Chain-of-Note [199]; Astute RAG [169]; Conflict diagnostics [15, 98]; OpenDecoder [118]
Fine-grained Attribution
FactScore [115]; ALCE [44]; GopherCite [113]; Self-RAG [6]; CoVe [31]; AlignScore [201]; MiniCheck [163]
Safety Evaluation
Prompt-injection defense [103]; StruQ [17]; RobustRAG [186]; CAR defense [182]
Utility Evaluation
RAGChecker [144]; RAGAS [37]; SePer [28]; RAGTruth [122]; FaithEval [116]; SafeRAG [95]; Robustness stress test [15]
Loop Orchestration
IRCoT [166]; ChainRAG [221]; LevelRAG [215]; MindSearch [20]; FLARE [73]; DRAGIN [161]; ReAct [193]; WebGPT [120]; TIR [160]
End-to-end Policy Learning
OpenRAG [219]; RouteRAG [51]; Search-R1 [75]; Toolformer [149]
Robust Retrieval (§3.2)
Context Assembly (§3.3)
Retrieval Verification
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
(§3.4)
Closed-Loop Training (§3.5)
Self-evolution
Reflexion [154]; MemGPT [128]; MemRL [214]; MemoryBank [218]; HippoRAG [53]; AutoRAG [80]; DSPy [78]
Figure 3: A multi-level denoising taxonomy aligned with the five-stage Section 3 pipeline: Controlled Indexing (§3.1), Robust Retrieval (§3.2), Context Assembly (§3.3), Retrieval Verification (§3.4), and Closed-Loop Training (§3.5). that avoid costly full rebuilds [64, 205], while STAR-RAG summarizes graph snapshots and T-GRAG attaches temporal edges to disambiguate overlapping versions at query time [87, 223]. The same principle extends to long-context memory, where timeline-based organization links retrieved episodes to their validity windows and curbs recency bias [125]. (4) Structure as a Defense to Noise. Flat indexing strips text of its relational and hierarchical context, leaving retrievers to operate over a bag of chunks; structural indexing restores this topology so that signal is preserved by construction. Graph-based methods replace arbitrary token windows with typed entity-relation networks—from GraphRAG and its variants supporting community-, subgraph-, and dual-level retrieval [36, 52, 59, 222], to G-Retriever,
which treats the graph itself as a first-class retrieval target [57], and HippoRAG, which uses personalized PageRank over an entity graph as a long-term memory substrate [53]. Because such structures only clarify when extraction is faithful, schema-aware canonicalization and retrieval-filtered construction sanitize the topology before use [203, 209]. Hierarchical indexing complements this by organizing evidence into recursive summary trees, enabling coarse-to-fine navigation while encapsulating local details within parent nodes for explicit source lineage [61, 148, 211].
3.2
Robust Retrieval
This subsection surveys interventions that reduce retriever-level noise before context assembly. While matching related documents
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
from a large corpus is already a denoising process, LLM-oriented retrieval further prioritizes precision and distractor resistance, as the inclusion of noise can severely degrade the generation quality of LLMs. We survey retrieval-time denoising along four dimensions: (i) query transformation, (ii) hybrid retrieval, (iii) distractor-aware retriever training, and (iv) fine-grained relevance matching. (1) Query Transformation. The mismatch between user intent and corpus representation is a starting point of retrieval noise. Raw queries are often underspecified or ambiguous, leading retrievers to fetch topically related but factually irrelevant candidates [25, 194]. To bridge this semantic gap, generative expansion methods utilize LLMs to generate hypothetical documents (e.g., HyDE) or pseudoanswers, projecting the query into a clearer document embedding space [43, 173, 198]; or abstract the query to a higher-level concept before retrieval to evoke broader supporting evidence [216]. For complex information needs, query decomposition breaks queries into simpler sub-queries, reducing the noise associated with singlestep retrieval over dense information [60, 110, 137]. Recent works further propose optimizing the rewriter via reinforcement learning (RL) or feedback loops, explicitly training rewriting to maximize end-to-end performance [14, 72, 111, 134]. (2) Hybrid Retrieval. Single retriever exhibit clear limitations: sparse methods overlook semantic nuances, while dense encoder are prone to semantic noise—retrieving topically related but factually irrelevant passages. Hybrid retrieval addresses this by fusing dense representations with sparse lexical signals (e.g., BM25), effectively using exact matching to ground semantic associations and filter hallucinations [11, 18, 39]. BGE-M3 further unify this fusion into a single backbone by jointly training dense, lexical, and multi-vector heads with a hybrid scoring rule [16]. Beyond static fusion, a line work dynamically invokes different retrieval method or LLM’s parametric memoery on demands to maximize the retrieval accuracy in different scenarios [76, 89, 109]. (3) Distractor-Aware Retriever Training. Enhancing the robustness of dense representations is critical for eliminating noise. Several axes strengthen the retriever training against distractors. Hard negative mining force the encoder to separate genuine evidence from topical near-misses [96, 140, 172], with adversarial frameworks dynamically escalating training difficulty, exposing the retriever to ever-subtler distractor [208]. Secondly, instruction tuning conditions the same embedding space on different task intents, so semantically similar but task-irrelevant passages are systematically downranked [27, 159, 177]. Finally, RAG-specific objectives further distill LLM-judged utility or defects back into the retriever, aligning what is retrieved with what the generator should actually exploit [97, 167]. (4) Fine-Grained Relevance Matching. Compressing document into a single vector inherently cause information loss and introducing noises [181]. Late interaction architectures like ColBERT retain token-level features for fine-grained matching, distinguishing subtle relevance from passages that are connected loosely [79, 146]. XTR scale this paragidm by streamlining training and scoring so that token-level retrieval no longer requires gather-and-rescore, cutting cost by orders of magnitude over ColBERT [83]. Methods like MUVERA directly train a multi-vector encoder to represent documents in a more fine-grained way beyond single-vector MIPS [32]. The same late-interaction principle has further been
Lu Dai et al.
ported to visually-rich documents via vision-language token interaction, broadening the scope beyond single-modality RAG [38].
3.3
Context Assembly
Context assembly is the first stage that take into account the interaction between retriever and generator, transforming a raw candidate set into an LLM prompt. The objective of this stage is to maximize information density under the constraints of LLM context windows. We categorize context assembly techniques into four denoising mechanisms: LLM-aware selection, prompt compression, layout optimization, and knowledge conflict resolving. (1) LLM-aware Selection and Pruning. After first-stage retrieval, reranking and pruning act as the first noise gate of context assembly, rigorously filtering irrelevant candidates before they enter the prompt. LLM-based reranking enhances pointwise cross-encoders [123] with holistic, set-level judgments: listwise permutation generation [107, 162], pairwise comparison [138], setwise most-relevant selection [224], and efficiency-oriented single-token decoding [142]. Diversity-aware subset selection moves beyond top-𝑘 truncation to balance relevance against redundancy via MMR [13] or Determinantal Point Processes [147]. Fine-grained context filtering operates at the sentence or token level to surgically remove non-supporting content. Speculative selection instructs the LLM itself to prune irrelevant evidence before reasoning [179]; trained pruners [23, 176] distill comparable signals into lightweight classifiers, and applies sentence-level reranking followed by LLM-based reconstruction of the surviving fragments [65] . Beyond rigid top-𝑘 paradigm, adaptive strategies can dynamically size the context based on query complexity [69] or abstain when evidence is insufficient [3, 112, 135]. (2) Long Context Compression. To mitigate noise from verbose or irrelevant tokens, compression techniques perform soft denoising. Extractive compression methods such as the LLMLingua family [70, 71, 130] use small models to score token informativeness and drop low-signal tokens. Abstractive compression synthesizes documents into concise meta-summaries or task-specific representations [91, 189]. A complementary line, soft-prompt compression, encodes retrieved passages into continuous embedding vectors—such as Gist Tokens [119], ICAE [45], and xRAG [22]—achieving extreme compression while preserving task performance. All three paradigms aim to maximize information gain per token [190], increasing the logical capacity of the context window. (3) Layout Optimization to Mitigate Position Bias. LLMs exhibit non-uniform attention distributions, often favoring the beginning and end of the context while neglecting the middle—the “lost in the middle” phenomenon [100]. Denoising in this dimension involves topology-aware arrangement, where retrieved passages are reordered to place the most critical evidence at attention peaks [136]. Rather than a naive concatenation based on retrieval scores, optimal layout strategies dynamically structure the prompt to align with the model’s inductive biases, ensuring that high-signal evidence is not statistically drowned out by surrounding context [151]. (4) Knowledge Conflict Resolution. Retrieved evidence is often internally inconsistent due to temporal drift, source disagreement, or duplicated paraphrases; naively flattening such passages into a single prompt can lead to hallucinations or reasoning paralysis [15, 98]. Conflict-aware assembly therefore treats contradiction as
LLM-Oriented Information Retrieval: A Denoising-First Perspective
a first-class denoising target: competing claims can be detected and clustered into explicit alternatives, or routed into a structured prompt that forces the model to adjudicate discrepancies rather than averaging them away [170, 187]. Complementarily, evidence grading methods such as Chain-of-Note [199] generate passagelevel reading notes and labels (e.g., supporting vs. irrelevant vs. contradictory), enabling the assembler to prune low-signal passages and to expose disagreements with clear provenance before final synthesis. OpenDecoder[118] incorporates retrieval relevance signals in answer decoding, modulating retrieval and generation with an internal attention network.
3.4
Retrieval Verification
Quality and fine-grained verification are essential for improving retrieval efficacy. They provide critical feedback for active denoising and establish an auditable chain of evidence for human review. In the LLM-oriented era, retrieval evaluation extends beyond traditional hit-based metrics to encompass utility, granular traceability, and security challenges. (1) Evaluating Utility of Retrieval. Effective denoising requires evaluating whether retrieved context genuinely benefits the generation, rather than merely relevance matching in an LLM-oriented system. Traditional metrics like NDCG fail to capture the nuanced relationship between retrieval and generation quality [26]. To disentangle these effects, LLM-as-a-judge frameworks such as RAGAS [37] and RAGChecker [144] separately measure components such as retrieval quality, context precision and answer faithfulness, making it possible to identify system failure modes. Furthermore, to isolate the true information gain, utility-based metrics like SePer [28] quantify the exact reduction in semantic uncertainty provided by the retrieval stage to measure retrieval utility. Complementary benchmarks such as RAGTruth [122], FaithEval [116], SafeRAG [95], and RGB [15] stress whether systems preserve faithfulness when evidence density degrades. (2) Fine-Grained Attribution of Long Answers. RAG systems often generate long-form answers that synthesize information across multiple retrieved passages, making it difficult for attribution. To localize hallucinations, recent works enforce fine-grained, span-level attribution rather than coarse document-level overlap. FactScore and AlignScore [115, 201] assess factuality by decomposing answers into atomic claims and evaluate as aligned pairs. Verifiable generation methods operationalize traceability by mandating quoted evidence [44, 113], using critic tokens during decoding [6], or applying post-hoc verification [31], with lightweight checkers [163] further make the evaluation efficiency computationally practical. (3) Retrieval Safety Auditing. The openness of LLM-oriented IR introduces adversarial “intentional noise” that demands verification beyond standard quality checks. Retrieved documents may contain indirect prompt injections or poisoned evidence designed to hijack the generation logic, override correct evidence, or corrupt long-term agent memory [21, 49, 103, 213, 225]. Against these threats, verification methods operate through complementary strategies: corroboration-based auditing such as CAR detects inconsistencies by seeking cross-source agreement under disinformation attacks [182]; isolation-and-aggregation frameworks such as
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
RobustRAG verify passage integrity and provide certifiable robustness guarantees against retrieval corruption [186]; and structuredquery constraints expose and limit instruction-channel contamination [17]. Security-focused benchmarks including SafeRAG [95] and RGB [15] systematically stress-test systems across diverse adversarial scenarios, transforming these verification outcomes into actionable denoising signals for upstream retrievers.
3.5
Closed-Loop Training
Search agents that interleave retrieval and reasoning offer capabilities beyond one-shot retrieval but are susceptible to unique vulnerabilities such as cumulative noise and error propagation. In multi-turn settings, a single irrelevant or misleading context can cascade into hallucinated reasoning, destabilizing the entire trajectory. Denoising thus shifts from static filtering to dynamic flow control. We investigate how agents mitigate noise through three mechanisms: adaptive loop orchestration, reinforcement learning, and self-evolution. Agentic loop orchestration and regulation. To mitigate cumulative noise, agents employ structured flow control as active noise filters. First, query decomposition minimizes semantic drift: methods like IRCoT [166] and ChainRAG [221] break complex queries into verifiable sub-steps to prevent “lost-in-retrieval” failures, while hierarchical planners in LevelRAG [215] and MindSearch [20] ensure narrowly targeted retrieval to minimize distractor intrusion. Second, adaptive scheduling acts as a dynamic gate: systems like Self-RAG [6], FLARE [73], and DRAGIN [161] assess parametric uncertainty to decide when to retrieve, blocking unnecessary contexts that would dilute focus. Finally, orchestration must incorporate adversarial filtering against prompt injection [103] and corpus poisoning [150, 225], requiring interface-level sanitization such as structured queries [17]. End-to-end training for RAG. Rather than relying on fixed heuristics, recent works optimize denoising directly for downstream utility. OpenRAG fine-tunes the retriever using generative feedback to filter distractors [219]. Crucially, reinforcement learning (RL) aligns the entire retrieval-generation loop: by optimizing against reward signals tied to output quality, RL equips LLMs to actively seek useful evidence and ignore retrieved noise [75, 210]. RouteRAG employs RL to learn a router policy about when to retrieve or rely on parametric knowledge [51]. Similarly, Toolformer learns to execute search APIs only when they reduce perplexity, avoiding calls that introduce context noise [149]. Denoising thus shifts from post-processing to a learned behavior encoded in model weights. Self-evolution and structural optimization. Beyond optimizing a fixed policy, agentic systems enhance denoising through continuous self-evolution. First, agents refine memory to prevent recurrent errors: Reflexion accumulates verbal feedback to bias trajectories away from noisy patterns [154], while MemGPT [128] and MemRL [214] implement explicit memory management—evicting irrelevant history or filtering “similar-but-useless” retrievals via learned utility—to maintain high signal-to-noise ratios. Second, at the system level, AutoRAG [80] and DSPy [78] treat the retrieval pipeline as a programmable surface, automatically searching for module combinations and prompt structures that maximize signal quality. These mechanisms transition denoising from a transient tactic to a
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
lifelong learning objective, where historical interaction and structural search drive continuous refinement of the agent’s information processing pipeline.
4
Practices in Retrieval-Augmented Applications
Whereas Section3 organizes denoising stage by stage, real retrievalaugmented applications fail through coupled noise that propagates across stages rather than isolated module errors. We illustrate this via four representative settings—coding agents, long-term memory assistants, deep research, and multimodal understanding—and summarize their failure signatures and interventions in Table 1.
4.1
Coding Agents
Coding agents operate in a distinct IR environment characterized by repository-scale search spaces and rigid execution constraints. Tasks in this domain, typified by benchmarks like SWE-bench [74, 126], require agents to navigate massive codebases to implement precise fixes. The central challenge is the extreme sparsity of actionable signals: a repository may contain millions of lines of code, yet the resolution to an issue often resides in a single function or a subtle cross-file dependency. Consequently, standard retrieval is insufficient; it is plagued by high-noise distractors such as distinct functions with identical names, deprecated utilities, and complex inheritance structures that mask the true root cause. To mitigate this, state-of-the-art workflows have evolved from flat retrieval-augmented generation into multi-stage pipelines that progressively filter noise through hierarchical localization, syntaxaware context management, and execution-based verification. First, systems like Agentless and RepoHyper employ a coarseto-fine filtering strategy [185] instead of retrieving code snippets directly. They typically start with keyword-based or vector search to identify candidate files, followed by a re-ranking or distinct selection phase where an LLM acts as a discriminator to prune irrelevant files based on the issue description. This hierarchical reduction transforms the retrieval problem into a decision-making process, ensuring that the limited attention budget of the downstream editing model is not diluted by structurally similar but logically unrelated code. Furthermore, to handle the "needle-in-a-haystack" nature of cross-file dependencies, retrieval is often augmented with static analysis graphs (e.g., call graphs or import dependencies), which serve as hard constraints to guide the agent toward logically connected components rather than merely textually similarity [33, 101]. Second, denoising bottleneck also lies in maximizing evidence density within the context window. Feeding raw source code is inefficient and noise-prone. Advanced agents address this by adopting syntax-aware representations, such as code skeletons with only class signatures and docstrings or AST-based slicing, which strip away implementation details to provide a high-level "map" of the repository [191, 204]. This technique, often referred to as a repository map or context compression, allows the LLM to maintain global awareness of project structure without being overwhelmed by local token noise. By presenting only the interface boundaries and hiding the method bodies until specifically requested, these systems artificially boost the signal-to-noise ratio, enabling the
Lu Dai et al.
model to reason about complex architectures that would otherwise exceed its context limits. Finally, coding agents uniquely employs execution and tests as a verification tool. The edit-execute loop allows agents to dynamically prune noisy trajectories using precise feedback from failing tests and stack traces. This signal enables systems like SWE-agent to distinguish plausible hallucinations from functional solutions, iteratively refining the retrieval context [4, 191]. Moreover, recent works suggest that strengthening this oracle via test augmentation or differential testing further prevents overfitting to sparse supervision [175, 196]. Consequently, retrieval pipeline in coding agents becomes a cyclic, verifiable loop where code structure and execution states continuously gate the flow of information.
4.2
Long-term Memory Assistants
Long-term memory assistants aim to maintain continuity across interaction horizons that far exceed a model’s context window. Typically, these systems externalize history into persistent storage, yet as interactions accumulate, the raw log data inevitably degrades into a high-entropy distribution characterized by redundancy, temporal obsolescence, and state conflict. In this setting, the accumulation of history paradoxically reduces retrieval precision, creating a noisy context that leads models to hallucinate based on superseded information [108, 128]. The primary noise source in lifelong interaction is temporal drift. Unlike static corpora, personal contexts are mutable; a user’s preference or status recorded at 𝑡 1 may be directly contradicted by an event at 𝑡 100 . Naive similarity-based retrieval often fails here, as outdated facts (“hard negatives”) share high semantic overlap with current queries, leading the LLM to hallucinate based on superseded evidence [108, 184]. Furthermore, indiscriminate logging creates a “store pollution” effect, where low-utility chitchat dilutes the retrieval pool, while adversarial inputs can introduce persistent poisoning risks that degrade reasoning across sessions [34, 225]. To mitigate these noise channels, denoising practices have shifted from passive indexing to active memory consolidation and structureaware retrieval. At the storage level, systems mimic human cognitive processes by abstracting raw episodic logs into synthesized semantic memories using reflection or recursive summarization to discard redundancy while preserving high-level traits [131, 218]. To combat ambiguity, unstructured text is often supplemented or replaced by structured representations, such as knowledge graphs or entity-attribute pairs, which enforce stricter schemas for state tracking and enable multi-hop reasoning over connected events [53, 63]. At the retrieval stage, standard dense retrieval is augmented with time-awareness via recency-weighted ranking or explicit validity filtering to ensure that the retrieved evidence reflects the current world state rather than historical artifacts [218]. Finally, during context construction, the focus lies on conflict resolution and verifiability. Rather than blindly packing top-𝑘 chunks, advanced pipelines apply deduplication and expose provenance metadata (e.g., timestamps, source attribution), allowing the LLM to adjudicate between conflicting memories and resist poisoning attacks [158, 200]. A shift is also happening in evaluation benchmarks, which have moved beyond simple fact recall to measuring robustness against
LLM-Oriented Information Retrieval: A Denoising-First Perspective
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
Table 1: Domain-aligned failure & recovery cases for denoising-first LLM-oriented IR (Sec. 4). Legend: Failure tags use C/R/A/V/L to denote the primary noise-entry layer (corpus/retriever/assembly/verification/loop): C1 provenance contamination; C2 redundancy/canonical drift; C3 temporal drift/obsolescence; R1 query/subquery drift; R2 hard distractors; A1 context dilution (lost-in-the-middle); A2 low information density (verbosity/redundancy); A3 evidence conflict (stale-state); V1 attribution gap; L1 cascading error propagation. Stage tags: CI controlled indexing, RR robust retrieval, CA context assembly, FV faithfulness verification, CL closed-loop improvement. Scenario
Failure Signature
Examples
Denoising Recipes
Key metrics
S4.1 Coding agents
R2 hard distractors (near-duplicate files/symbols); L1 early mistake locks an incorrect plan; C2 stale/deprecated code index. Symptom: plausible patch, wrong target.
SWE-bench-style issue: FAIL_TO_PASS points to src/core/parser.py; top-𝑘 retrieval ranks src/legacy/parser.py higher; agent patches legacy → tests still fail.
RR BM25+dense+symbol-aware retrieval; hard-negative rerank; CI syntax-aware repo indexing; CA de-dup clones; minimal evidence (failing trace + call graph). FV sandbox tests; reject mis-targeted patches. CL train with distractive counterexamples.
TASK % Resolved [126]; Pass Rate [114]; EFF n_turns, n_toolcalls, n_total_tokens; time_to_last_token. TASK ruff/mypy/bandit pass. JUDGE Agentic Rubrics [141]. EFF Cost per Task ($); LOC file/function localization accuracy.
S4.2 Long-term memory assistant
C2 memory store pollution
2024-03: “I live in Boston.” 2024-06: “I moved to Seattle.” Later: “recommend food near me” → Boston suggestions (stale memory wins).
CI time-aware indexing; CI memory consolidation; RR recency-aware rerank + time-range filtering/query expansion. FV state-consistency check (detect contradictions; ask/abstain).
TASK Accuracy [184]; TASK Answer Prediction (F1) [108]; BLEU/ROUGE; FactScore [115]; RETR Recall@K/NDCG@K. JUDGE Faithfulness & Context Relevance [37]. JUDGE RACE (Overall, Comprehensiveness, Depth, Instruction-Following, Readability). CITE FACT (Citation Accuracy,Effective Citation) [35]; UTILITY SePer [28]. CITE Citation Precision/Recall [44].
C3 temporal drift; A3
stale-state conflict across memories. Symptom: response reflects outdated preference/profile. S4.3 Deep research and reports
R1 subquery drift; A1 evidence dilution; V1 attribution gap. Symptom: confident claims with weak/irrelevant citations.
Query: “Compare PFAS regulations US vs EU (2025).” Subqueries drift to general health-effects; report claims “banned globally” with a citation to an unrelated blog snippet.
RR subquery anchoring + novelty/duplicate filter. CA claim-level evidence packing (“cite-by-claim”). FV entailment check between each claim and cited spans. CL RL-trained search policies optimize evidence density.
S4.4 Multimodal understanding
A2 event sparsity in long contexts; V1 weak timestamp grounding. Symptom: misses key moment or hallucinates timestamps.
45-min lecture video: “When is Theorem 2 stated?” Retrieval picks a 02:10 subtitle match; ground truth is 37:48 → wrong timestamp.
CI multimodal segment indexing. RR TASK Accuracy; F1 score; RETR dual-channel retrieval; fine-grained Recall [40, 42, 220]; LOC mAP (IoU=0.5/0.7) & HIT@1 [85]; temporal rerank. CA event-centric segment pooling/keyframe selection. TASK MAE & MSE & RMSE [139]; FV timestamp verification loop and JUDGE MM-Relevance [108] re-query.
drift and inconsistency. Datasets like LoCoMo [108] and LongMemEval [184] explicitly probe an agent’s ability to ignore stale confounders and update beliefs in response to new evidence, underscoring that the utility of long-term memory depends less on how much is stored, and more on how effectively noise is suppressed.
4.3
Deep Research
Deep research systems operate over open-ended queries, heterogeneous corpora, and evolving intermediate hypotheses. Typically centered on an LLM planner and given a broad query Q, a deep research system: (i) decomposes Q into a structured plan 𝑃 = 𝑔1, . . . , 𝑔𝑛 of subgoals, (ii) executes multi-hop retrieval over external and internal corpora to collect evidence sets 𝑅𝑖 for each 𝑔𝑖 , (iii) iteratively synthesizes intermediate products while adapting the plan via dynamic replanning and meta-cognitive validation, and (iv) assembles a structured artifact with explicit citations and provenance [62, 68, 153]. Unlike short-form QA, failures emerge from compounding noise across planning, retrieval, context construction, and verification, exhausting the model’s attention and utilization capacity [35, 104, 156].
Noise in deep research is multi-stage and structural. At the access layer, iterative sub-query generation is prone to semantic drift: vague sub-queries retrieve topically adjacent but irrelevant literature, diluting evidence pools [2, 58]. Even expressive LLM-based retrievers underperform due to query collapse into shallow keyword patterns, amplifying retrieval noise [58]. Domain-specific restriction and metadata augmentation partially mitigate this, indicating that denoising pressure can shift to the data layer [5]. Context construction constitutes another bottleneck. Retrieved documents are typically long, redundant, and internally noisy; naïvely concatenating them leads to positional bias and attention collapse [100]. Multiple systems therefore introduce summarization, filtering, or re-ranking stages that transform raw documents into query-conditioned evidence snippets before insertion into the reasoning context, trading recall for evidence density and enabling agents to examine orders of magnitude more text than fits into the prompt [90, 156]. Beyond soft filtering, recent work emphasizes hard denoising mechanisms. Neuro-symbolic slicing methods formalize denoising as span selection, predicting explicit indices and discarding all other
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
content prior to reasoning [104]. This reframes the dominant failure mode as a retrieval–utilization gap: even when gold evidence is retrieved, it competes with overwhelming distractors. By making exclusion explicit, such methods expose utilization as a first-class bottleneck in deep research pipelines. Verification and post-generation feedback introduce a critical denoising stage. Long-form reports are vulnerable to unsupported claims, citation hallucinations, and internal contradictions. Systems therefore incorporate self-review or multi-agent verification loops, auditing drafts for citation accuracy and factual consistency [5, 47, 48]. Benchmarks increasingly penalize spurious or missing citations, while revealing that even when retrieval succeeds, agents struggle to organize evidence into expert-like structures [35, 212]. Architecturally, deep research systems differ in where they allocate denoising effort. Planner-centric approaches reduce noise upstream by constraining subgoals and search trajectories, while draft-centric or diffusion-style systems treat the report as a noisy latent variable iteratively refined through targeted retrieval and revision [54]. Reinforcement-learning-based agents further internalize denoising by coupling outcome rewards with evidence-density or process-level signals, encouraging policies that retrieve less but better information [56, 75, 217]. Across these variants, improved performance is consistently associated with explicit mechanisms for pruning, summarizing, or verifying evidence, rather than with raw increases in retrieval breadth.
4.4
Multimodal Understanding
The transition from static text to multimodal temporal data—spanning time series, genomics, and video—introduces temporal redundancy. Temporal streams scale linearly with duration 𝑇 , while causal evidence 𝐸𝑐𝑎𝑢𝑠𝑎𝑙 ⊂ 𝑇 remains sparse. Thus, the retrieval challenge shifts from indexing documents to preserving causal signals within continuous, high-entropy streams. In spatio-temporal modeling, naive context window expansion often degrades performance due to distribution shifts and nonstationary noise [165], shifting the objective from maximizing context length to optimizing evidence density. Recent architectures treat history selection as a denoising operator: RATD [99] and TSRAG [121] replace fixed-window horizons with dynamic retrieval that fetches only segments maximizing predictive utility, while RAFT [55] retrieves training samples exhibiting similar temporal dynamics. To decouple short-term volatility from long-term trends, Time-MoE [152] routes tokens to experts specialized in specific temporal resolutions, and dual-masking strategies filter high-entropy timesteps during training to prevent overfitting to noise [41]. This sparsity is more pronounced in biological and visual domains. In genomics, models like Gene42 [168] scale attention to extreme lengths (e.g., 192k tokens), yet must discriminate functional regulatory motifs from vast non-coding regions. Video data represents the lower bound of signal-to-noise ratio, with relevant evidence often occupying less than 1% of frames. The Video-MME benchmark [40] quantifies a "long-context gap," where performance inversely correlates with duration, indicating that unpruned context acts as pollution.
Lu Dai et al.
To address this, architectures are evolving into dual-channel retrieval systems separating semantic reasoning from evidence localization. VideoRAG [105] exemplifies a denoising-first strategy, using Knowledge Graphs for semantic retrieval while limiting dense visual matching to relevant intervals. Complementary approaches leverage low-noise modalities (e.g., ASR transcripts) as anchors to index high-entropy visual streams. Hierarchical systems like VideoTree [180] introduce query-adaptive pruning, filtering irrelevant branches before they reach reasoning, acting as a coarse-to-fine noise filter. Finally, verifiability in multimodal generation requires explicit grounding. To prevent hallucination in temporal localization, models like VTG-LLM [50] and TimeExpert [192] decouple timestamp generation from captioning via specialized experts. By mandating timestamped citations (e.g., 𝑡𝑠𝑡𝑎𝑟𝑡 , 𝑡𝑒𝑛𝑑 ), these methods transform generation into provenance-aware output, ensuring visual evidence is causally linked to the generated reasoning.
5
Future Directions
We argue that LLM-oriented IR must evolve from a passive retrieval utility into an active programmable noise gate. The core objective shifts from maximizing isolated recall to maximizing usable evidence density within the model’s cognitive budget. (1) Utility-Centric Evaluation. Standard ranking metrics often diverge from generation quality. Future benchmarks must measure causal utility—rewarding retrieval only when it resolves reasoning gaps or corrects hallucinations—and penalize unverified attribution, enforcing a strict evidence-to-generation contract. (2) Proactive Index Sanitation. To combat the “pollution” of synthetic and stale content, indexing must transition to stratified governance. Systems should treat provenance, temporal validity, and cryptographic signatures (e.g., C2PA) as hard constraints, filtering low-credibility signals before they compete for attention. (3) Self-Evolving Retrieval Loops. Static retrievers degrade in dynamic environments. We envision agents that employ closed-loop feedback to refine search policies in real-time. By learning from downstream reasoning failures, the retriever adapts to filter hard distractors and prevent “spiral” errors, turning rejection signals into upstream optimization. (4) Optimizing Information Density in Context. Addressing context limitations requires moving beyond document concatenation. Future context assembly should operate on atomic evidence units (e.g., specific claims, code symbols) rather than passages. This structural compression maximizes the logical capacity of the prompt, ensuring every token contributes to verifiability.
6
Acknowledgements
This work was supported in part by the National Key R&D Program of China (Grant No.2023YFF0725001), in part by the National Natural Science Foundation of China (Grant No. 62572417, Grant No.92370204), in part by the Guangdong Basic and Applied Basic Research Foundation (Grant No.2023B1515120057), in part by the Key-Area Special Project of Guangdong Provincial Ordinary Universities (Grant No.2024ZDZX1007).
LLM-Oriented Information Retrieval: A Denoising-First Perspective
References [1] Amro Abbas, Kushal Tirumala, Dániel Simig, Surya Ganguli, and Ari S Morcos. 2024. SemDeDup: Data-efficient Learning at Web-scale through Semantic Deduplication. In ICLR Workshop on Multimodal Representation Learning. [2] Shubham Agarwal, Gaurav Sahu, Abhay Puri, Issam H Laradji, Krishnamurthy DJ Dvijotham, Jason Stanley, Laurent Charlin, and Christopher Pal. 2024. Litllm: A toolkit for scientific literature review. arXiv preprint arXiv:2402.01788 (2024). [3] Chen Amiraz, Florin Cuconasu, Simone Filice, and Zohar Karnin. 2025. The distracting effect: Understanding irrelevant passages in rag. arXiv preprint arXiv:2505.06914 (2025). [4] Antonis Antoniades, Albert Örwall, Kexun Zhang, Yuxi Xie, Anirudh Goyal, and William Yang Wang. 2025. SWE-Search: Enhancing Software Agents with Monte Carlo Tree Search and Iterative Refinement. In The Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025. OpenReview.net. https://openreview.net/forum?id=G7sIFXugTX [5] Akari Asai, Jacqueline He, Rulin Shao, Weijia Shi, Amanpreet Singh, Joseph Chee Chang, Kyle Lo, Luca Soldaini, Sergey Feldman, Mike D’Arcy, et al. 2026. Synthesizing scientific literature with retrieval-augmented language models. Nature (2026), 1–7. [6] Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. [n. d.]. Self-RAG: Learning to Retrieve, Generate, and Critique through SelfReflection. In The Twelfth International Conference on Learning Representations. [7] Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, et al. 2024. Longbench: A bilingual, multitask benchmark for long context understanding. In Proceedings of the 62nd annual meeting of the association for computational linguistics (volume 1: Long papers). 3119–3137. [8] Sylvio Barbon Junior, Paolo Ceravolo, Sven Groppe, Mustafa Jarrar, Samira Maghool, Florence Sèdes, Soror Sahri, and Maurice Van Keulen. 2024. Are large language models the new interface for data pipelines?. In Proceedings of the International Workshop on Big Data in Emergent Distributed Environments. 1–6. [9] Sergey Brin and Lawrence Page. 1998. The anatomy of a large-scale hypertextual web search engine. Computer networks and ISDN systems 30, 1-7 (1998), 107–117. [10] Andrei Z Broder. 1997. On the resemblance and containment of documents. In Proceedings. Compression and Complexity of SEQUENCES 1997 (Cat. No. 97TB100171). IEEE, 21–29. [11] Sebastian Bruch, Siyu Gai, and Amir Ingber. 2023. An analysis of fusion functions for hybrid retrieval. ACM Transactions on Information Systems 42, 1 (2023), 1–35. [12] Vannevar Bush et al. 1945. As we may think. The atlantic monthly 176, 1 (1945), 101–108. [13] Jaime G. Carbonell and Jade Goldstein. 1998. The Use of MMR, DiversityBased Reranking for Reordering Documents and Producing Summaries. In SIGIR ’98: Proceedings of the 21st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, August 24-28 1998, Melbourne, Australia, W. Bruce Croft, Alistair Moffat, C. J. van Rijsbergen, Ross Wilkinson, and Justin Zobel (Eds.). ACM, 335–336. doi:10.1145/290941.291025 [14] Chi-Min Chan, Chunpu Xu, Ruibin Yuan, Hongyin Luo, Wei Xue, Yike Guo, and Jie Fu. [n. d.]. RQ-RAG: Learning to Refine Queries for Retrieval Augmented Generation. In First Conference on Language Modeling. [15] Jiawei Chen, Hongyu Lin, Xianpei Han, and Le Sun. 2024. Benchmarking large language models in retrieval-augmented generation. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 17754–17762. [16] Jianlyu Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024. M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. In Findings of the association for computational linguistics: ACL 2024. 2318–2335. [17] Sizhe Chen, Julien Piet, Chawin Sitawarin, and David Wagner. 2025. { StruQ } : Defending against prompt injection with structured queries. In 34th USENIX Security Symposium (USENIX Security 25). 2383–2400. [18] Tao Chen, Mingyang Zhang, Jing Lu, Michael Bendersky, and Marc Najork. 2022. Out-of-domain semantics to the rescue! zero-shot hybrid retrieval models. In European Conference on Information Retrieval. Springer, 95–110. [19] Xiaoyang Chen, Ben He, Hongyu Lin, Xianpei Han, Tianshu Wang, Boxi Cao, Le Sun, and Yingfei Sun. 2024. Spiral of Silence: How is Large Language Model Killing Information Retrieval?—A Case Study on Open Domain Question Answering. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 14930–14951. [20] Zehui Chen, Kuikun Liu, Qiuchen Wang, Jiangning Liu, Wenwei Zhang, Kai Chen, and Feng Zhao. [n. d.]. MindSearch: Mimicking Human Minds Elicits Deep AI Searcher. In The Thirteenth International Conference on Learning Representations. [21] Zhaorun Chen, Zhen Xiang, Chaowei Xiao, Dawn Song, and Bo Li. 2024. Agentpoison: Red-teaming llm agents via poisoning memory or knowledge bases. Advances in Neural Information Processing Systems 37, 130185–130213.
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
[22] Xin Cheng, Xun Wang, Xingxing Zhang, Tao Ge, Si-Qing Chen, Furu Wei, Huishuai Zhang, and Dongyan Zhao. 2024. xrag: Extreme context compression for retrieval-augmented generation with one token. Advances in Neural Information Processing Systems 37, 109487–109516. [23] Nadezhda Chirkova, Thibault Formal, Vassilina Nikoulina, and Stéphane CLINCHANT. [n. d.]. Provence: efficient and robust context pruning for retrievalaugmented generation. In The Thirteenth International Conference on Learning Representations. [24] Content Credentials. 2025. C2PA Technical Specification v2. 2. [25] W Bruce Croft, Donald Metzler, Trevor Strohman, et al. 2010. Search engines: Information retrieval in practice. Vol. 520. Addison-Wesley Reading. [26] Florin Cuconasu, Giovanni Trappolini, Federico Siciliano, Simone Filice, Cesare Campagnano, Yoelle Maarek, Nicola Tonellotto, and Fabrizio Silvestri. 2024. The Power of Noise: Redefining Retrieval for RAG Systems. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 719–729. [27] Lu Dai, Hao Liu, and Hui Xiong. 2024. Improve Dense Passage Retrieval with Entailment Tuning. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 11375–11387. [28] Lu Dai, Yijie Xu, Jinhui Ye, Hao Liu, and Hui Xiong. [n. d.]. SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction. In The Thirteenth International Conference on Learning Representations. [29] Sunhao Dai, Weihao Liu, Yuqi Zhou, Liang Pang, Rongju Ruan, Gang Wang, Zhenhua Dong, Jun Xu, and Ji-Rong Wen. 2024. Cocktail: A Comprehensive Information Retrieval Benchmark with LLM-Generated Documents Integration. In Findings of the Association for Computational Linguistics ACL 2024. 7052–7074. [30] Sumanth Dathathri, Abigail See, Sumedh Ghaisas, Po-Sen Huang, Rob McAdam, Johannes Welbl, Vandana Bachani, Alex Kaskasoli, Robert Stanforth, Tatiana Matejovicova, et al. 2024. Scalable watermarking for identifying large language model outputs. Nature 634, 8035 (2024), 818–823. [31] Shehzaad Dhuliawala, Mojtaba Komeili, Jing Xu, Roberta Raileanu, Xian Li, Asli Celikyilmaz, and Jason E Weston. 2024. Chain-of-Verification Reduces Hallucination in Large Language Models. In ICLR 2024 Workshop on Reliable and Responsible Foundation Models. [32] Laxman Dhulipala, Majid Hadian, Rajesh Jayaram, Jason Lee, and Vahab Mirrokni. 2024. Muvera: Multi-vector retrieval via fixed dimensional encoding. Advances in Neural Information Processing Systems 37, 101042–101073. [33] Yangruibo Ding, Zijian Wang, Wasi Uddin Ahmad, Hantian Ding, Ming Tan, Nihal Jain, Murali Krishna Ramanathan, Ramesh Nallapati, Parminder Bhatia, Dan Roth, and Bing Xiang. 2023. CrossCodeEval: A Diverse and Multilingual Benchmark for Cross-File Code Completion. In NeurIPS. arXiv:2310.11248 [cs.LG] http://arxiv.org/abs/2310.11248v2 arXiv:2310.11248. [34] Shen Dong, Shaochen Xu, Pengfei He, Yige Li, Jiliang Tang, Tianming Liu, Hui Liu, and Zhen Xiang. 2025. Memory Injection Attacks on LLM Agents via Query-Only Interaction. arXiv:2503.03704 [cs.LG] [35] Mingxuan Du, Benfeng Xu, Chiwei Zhu, Xiaorui Wang, and Zhendong Mao. 2025. DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents. arXiv preprint arXiv:2506.11763 (2025). [36] Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, Dasha Metropolitansky, Robert Osazuwa Ness, and Jonathan Larson. 2024. From local to global: A graph rag approach to queryfocused summarization. arXiv preprint arXiv:2404.16130 (2024). [37] Shahul ES, Jithin James, Luis Espinosa Anke, and Steven Schockaert. 2023. RAGAs: Automated Evaluation of Retrieval Augmented Generation. Conference of the European Chapter of the Association for Computational Linguistics abs/2309.15217 (2023). [38] Manuel Faysse, Hugues Sibille, Tony Wu, Bilel Omrani, Gautier Viaud, CELINE HUDELOT, and Pierre Colombo. [n. d.]. ColPali: Efficient Document Retrieval with Vision Language Models. In The Thirteenth International Conference on Learning Representations. [39] Thibault Formal, Benjamin Piwowarski, and Stéphane Clinchant. 2021. SPLADE: Sparse lexical and expansion model for first stage ranking. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2288–2292. [40] Chaoyou Fu, Yuhan Dai, Yongdong Luo, Lei Li, Shuhuai Ren, Renrui Zhang, Zihan Wang, Chenyu Zhou, Yunhang Shen, Mengdan Zhang, Peixian Chen, Yanwei Li, Shaohui Lin, Sirui Zhao, Ke Li, Tong Xu, Xiawu Zheng, Enhong Chen, Caifeng Shan, Ran He, and Xing Sun. 2025. Video-mme: The first-ever comprehensive evaluation benchmark of multi-modal llms in video analysis. In CVPR. [41] Yisong Fu, Zezhi Shao, Chengqing Yu, Yujie Li, Zhulin An, Cheems Wang, Yongjun Xu, and Fei Wang. 2025. Selective Learning for Deep Time Series Forecasting. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=kgzRy6nD6D [42] Jiyang Gao, Chen Sun, Zhenheng Yang, and Ram Nevatia. 2017. Tall: Temporal activity localization via language query. In Proceedings of the IEEE international conference on computer vision. 5267–5275.
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
[43] Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2023. Precise Zero-Shot Dense Retrieval without Relevance Labels.. In Annual Meeting of the Association for Computational Linguistics (ACL). 1762–1777. [44] Tianyu Gao, Howard Yen, Jiatong Yu, and Danqi Chen. 2023. Enabling large language models to generate text with citations. arXiv preprint arXiv:2305.14627 (2023). [45] Tao Ge, Hu Jing, Lei Wang, Xun Wang, Si-Qing Chen, and Furu Wei. [n. d.]. In-context Autoencoder for Context Compression in a Large Language Model. In The Twelfth International Conference on Learning Representations. [46] Sebastian Gehrmann, Hendrik Strobelt, and Alexander M Rush. 2019. GLTR: Statistical Detection and Visualization of Generated Text. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: System Demonstrations. 111–116. [47] Gregory Hok Tjoan Go, Khang Ly, Anders Søgaard, Amin Tabatabaei, Maarten de Rijke, and Xinyi Chen. 2025. LiRA: A Multi-Agent Framework for Reliable and Readable Literature Review Generation. arXiv preprint arXiv:2510.05138 (2025). [48] Alon Gorenshtein, Kamel Shihada, Moran Sorka, Dvir Aran, and Shahar Shelly. 2025. LITERAS: Biomedical literature review and citation retrieval agents. Computers in Biology and Medicine 192 (2025), 110363. [49] Kai Greshake, Sahar Abdelnabi, Shailesh Mishra, Christoph Endres, Thorsten Holz, and Mario Fritz. 2023. Not what you’ve signed up for: Compromising realworld llm-integrated applications with indirect prompt injection. In Proceedings of the 16th ACM workshop on artificial intelligence and security. 79–90. [50] Yongxin Guo, Jingyu Liu, Mingda Li, Dingxin Cheng, Xiaoying Tang, Dianbo Sui, Qingbin Liu, Xi Chen, and Kevin Zhao. 2025. Vtg-llm: Integrating timestamp knowledge into video llms for enhanced video temporal grounding. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 3302–3310. [51] Yucan Guo, Miao Su, Saiping Guan, Zihao Sun, Xiaolong Jin, Jiafeng Guo, and Xueqi Cheng. 2025. RouteRAG: Efficient Retrieval-Augmented Generation from Text and Graph via Reinforcement Learning. arXiv preprint arXiv:2512.09487 (2025). [52] Zirui Guo, Lianghao Xia, Yanhua Yu, Tu Ao, and Chao Huang. 2024. Lightrag: Simple and fast retrieval-augmented generation. arXiv preprint arXiv:2410.05779 (2024). [53] Bernal Jiménez Gutiérrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. 2024. HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models. In NeurIPS. arXiv:2405.14831 [cs.CL] NeurIPS 2024 (per arXiv comments). [54] Rujun Han, Yanfei Chen, Zoey CuiZhu, Lesly Miculicich, Guan Sun, Yuanjun Bi, Weiming Wen, Hui Wan, Chunfeng Wen, Solène Maître, et al. 2025. Deep researcher with test-time diffusion. arXiv preprint arXiv:2507.16075 (2025). [55] Sungwon Han, Seungeon Lee, Meeyoung Cha, Sercan O Arik, and Jinsung Yoon. 2025. Retrieval Augmented Time Series Forecasting. In Forty-second International Conference on Machine Learning. [56] Bowei He, Minda Hu, Zenan Xu, Hongru Wang, Licheng Zong, Yankai Chen, Chen Ma, Xue Liu, Pluto Zhou, and Irwin King. 2026. Search-R2: Enhancing Search-Integrated Reasoning via Actor-Refiner Collaboration. arXiv preprint arXiv:2602.03647 (2026). [57] Xiaoxin He, Yijun Tian, Yifei Sun, Nitesh Chawla, Thomas Laurent, Yann LeCun, Xavier Bresson, and Bryan Hooi. 2024. G-retriever: Retrieval-augmented generation for textual graph understanding and question answering. Advances in Neural Information Processing Systems 37 (2024), 132876–132907. [58] Tiansheng Hu, Yilun Zhao, Canyu Zhang, Arman Cohan, and Chen Zhao. 2026. SAGE: Benchmarking and Improving Retrieval for Deep Research Agents. arXiv preprint arXiv:2602.05975 (2026). [59] Yuntong Hu, Zhihan Lei, Zheng Zhang, Bo Pan, Chen Ling, and Liang Zhao. 2024. GRAG: Graph Retrieval-Augmented Generation. arXiv preprint arXiv:2405.16506 (2024). [60] Chao-Wei Huang, Chen-Yu Hsu, Tsu-Yuan Hsu, Chen-An Li, and Yun-Nung Chen. 2023. CONVERSER: Few-shot Conversational Dense Retrieval with Synthetic Data Generation.. In SIGdial Meetings (SIGDIAL). 381–387. [61] Haoyu Huang, Yongfeng Huang, Junjie Yang, Zhenyu Pan, Yongchao Chen, Kaixiong Ma, Hongzhi Chen, and Jiawei Cheng. 2025. Retrieval-Augmented Generation with Hierarchical Knowledge. In Findings of the Association for Computational Linguistics: EMNLP 2025. [62] Yuxuan Huang, Yihang Chen, Haozheng Zhang, Kang Li, Huichi Zhou, Meng Fang, Linyi Yang, Xiaoguang Li, Lifeng Shang, Songcen Xu, et al. 2025. Deep research agents: A systematic examination and roadmap. arXiv preprint arXiv:2506.18096 (2025). [63] Zhengjun Huang, Zhoujin Tian, Qintian Guo, Fangyuan Zhang, Yingli Zhou, Di Jiang, Zeying Xie, and Xiaofang Zhou. 2025. LiCoMemory: Lightweight and Cognitive Agentic Memory for Efficient Long-Term Reasoning. arXiv:2511.01448 [cs.IR] [64] Daniel Huwiler, Kurt Stockinger, and Jonathan Fürst. 2025. VersionRAG: VersionAware Retrieval-Augmented Generation for Evolving Documents. arXiv preprint arXiv:2510.08109 (2025).
Lu Dai et al.
[65] Taeho Hwang, Soyeong Jeong, Sukmin Cho, SeungYoon Han, and Jong C Park. 2024. DSLR: Document refinement with sentence-level re-ranking and reconstruction to enhance retrieval-augmented generation. In Proceedings of the 3rd Workshop on Knowledge Augmented Methods for NLP. 73–92. [66] Gautier Izacard, Mathilde Caron, Lucas Hosseini, Sebastian Riedel, Piotr Bojanowski, Armand Joulin, and Edouard Grave. [n. d.]. Unsupervised Dense Information Retrieval with Contrastive Learning. Transactions on Machine Learning Research ([n. d.]). [67] Gautier Izacard and Edouard Grave. 2021. Leveraging passage retrieval with generative models for open domain question answering. In Proceedings of the 16th conference of the european chapter of the association for computational linguistics: main volume. 874–880. [68] Abhinav Java, Ashmit Khandelwal, Sukruta Midigeshi, Aaron Halfaker, Amit Deshpande, Navin Goyal, Ankur Gupta, Nagarajan Natarajan, and Amit Sharma. 2025. Characterizing deep research: A benchmark and formal definition. arXiv preprint arXiv:2508.04183 (2025). [69] Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, and Jong C Park. 2024. Adaptive-rag: Learning to adapt retrieval-augmented large language models through question complexity. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 7036–7050. [70] Huiqiang Jiang, Qianhui Wu, Chin-Yew Lin, Yuqing Yang, and Lili Qiu. 2023. Llmlingua: Compressing prompts for accelerated inference of large language models. arXiv preprint arXiv:2310.05736 (2023). [71] Huiqiang Jiang, Qianhui Wu, Xufang Luo, Dongsheng Li, Chin-Yew Lin, Yuqing Yang, and Lili Qiu. 2024. Longllmlingua: Accelerating and enhancing llms in long context scenarios via prompt compression. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 1658–1677. [72] Pengcheng Jiang, Jiacheng Lin, Lang Cao, Runchu Tian, SeongKu Kang, Zifeng Wang, Jimeng Sun, and Jiawei Han. 2025. DeepRetrieval: Hacking Real Search Engines and Retrievers with Large Language Models via Reinforcement Learning. arXiv.org abs/2503.00223 (2025). [73] Zhengbao Jiang, Frank F Xu, Luyu Gao, Zhiqing Sun, Qian Liu, Jane Dwivedi-Yu, Yiming Yang, Jamie Callan, and Graham Neubig. 2023. Active retrieval augmented generation. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 7969–7992. [74] Carlos E. Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik R. Narasimhan. 2024. SWE-bench: Can Language Models Resolve Real-world Github Issues?. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net. https://openreview.net/forum?id=VTF8yNQM66 [75] Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. 2025. Search-r1: Training llms to reason and leverage search engines with reinforcement learning. arXiv preprint arXiv:2503.09516 (2025). [76] Jushaan Singh Kalra, Xinran Zhao, To Eun Kim, Fengyu Cai, Fernando Diaz, and Tongshuang Wu. 2025. MoR: Better Handling Diverse Queries with a Mixture of Sparse, Dense, and Human Retrievers. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 11982–12001. [77] Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick SH Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense Passage Retrieval for Open-Domain Question Answering.. In EMNLP (1). 6769–6781. [78] Omar Khattab, Arnav Singhvi, Paridhi Maheshwari, Zhiyuan Zhang, Keshav Santhanam, Saiful Haq, Ashutosh Sharma, Thomas T Joshi, Hanna Moazam, Heather Miller, et al. 2024. Dspy: Compiling declarative language model calls into state-of-the-art pipelines. In The Twelfth International Conference on Learning Representations. [79] Omar Khattab and Matei Zaharia. 2020. Colbert: Efficient and effective passage search via contextualized late interaction over bert. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval. 39–48. [80] Dongkyu Kim, Byoungwook Kim, Donggeon Han, and Matouš Eibich. 2024. Autorag: Automated framework for optimization of retrieval augmented generation pipeline. arXiv preprint arXiv:2410.20878 (2024). [81] John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. 2023. A watermark for large language models. In International Conference on Machine Learning. PMLR, 17061–17084. [82] Yuan-Ching Kuo, Yi Yu, Chih-Ming Chen, and Chuan-Ju Wang. 2025. MMLF: Multi-query Multi-passage Late Fusion Retrieval. In Findings of the Association for Computational Linguistics: NAACL 2025. 6587–6598. [83] Jinhyuk Lee, Zhuyun Dai, Sai Meher Karthik Duddu, Tao Lei, Iftekhar Naim, Ming-Wei Chang, and Vincent Zhao. 2023. Rethinking the role of token retrieval in multi-vector retrieval. Advances in Neural Information Processing Systems 36, 15384–15405. [84] Katherine Lee, Daphne Ippolito, Andrew Nystrom, Chiyuan Zhang, Douglas Eck, Chris Callison-Burch, and Nicholas Carlini. 2022. Deduplicating training data makes language models better. In Proceedings of the 60th Annual Meeting of the
LLM-Oriented Information Retrieval: A Denoising-First Perspective
Association for Computational Linguistics (Volume 1: Long Papers). 8424–8445. [85] Jie Lei, Tamara L Berg, and Mohit Bansal. 2021. Detecting moments and highlights in videos via natural language queries. Advances in Neural Information Processing Systems 34 (2021), 11846–11858. [86] Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems 33 (2020), 9459–9474. [87] Dong Li, Yichen Niu, Ying Ai, Xiang Zou, Biqing Qi, and Jianxing Liu. 2025. T-grag: A dynamic graphrag framework for resolving temporal conflicts and redundancy in knowledge retrieval. In Proceedings of the 33rd ACM International Conference on Multimedia. 11880–11889. [88] Jeffrey Li, Alex Fang, Georgios Smyrnis, Maor Ivgi, Matt Jordan, Samir Yitzhak Gadre, Hritik Bansal, Etash Guha, Sedrick Keh, Kushal Arora, et al. 2024. DataComp-LM: In Search of the Next Generation of Training Sets for Language Models. In Advances in Neural Information Processing Systems. [89] Millicent Li, Tongfei Chen, Benjamin Van Durme, and Patrick Xia. 2025. MultiField Adaptive Retrieval. In The Thirteenth International Conference on Learning Representations. [90] Xiaoxi Li, Guanting Dong, Jiajie Jin, Yuyao Zhang, Yujia Zhou, Yutao Zhu, Peitian Zhang, and Zhicheng Dou. 2025. Search-o1: Agentic search-enhanced large reasoning models. arXiv preprint arXiv:2501.05366 (2025). [91] Yucheng Li, Bo Dong, Frank Guerin, and Chenghua Lin. 2023. Compressing context to enhance inference efficiency of large language models. In Proceedings of the 2023 conference on empirical methods in natural language processing. 6342– 6353. [92] Yafu Li, Qintong Li, Leyang Cui, Wei Bi, Zhilin Wang, Longyue Wang, Linyi Yang, Shuming Shi, and Yue Zhang. 2024. MAGE: Machine-generated text detection in the wild. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 36–53. [93] Zhicong Li, Jiahao Wang, Zhishu Jiang, Hangyu Mao, Zhongxia Chen, Jiazhen Du, Yuanxing Zhang, Fuzheng Zhang, Di Zhang, and Yong Liu. 2024. Dmqr-rag: Diverse multi-query rewriting for rag. arXiv preprint arXiv:2411.13154 (2024). [94] Weixin Liang, Yaohui Zhang, Mihai Codreanu, Jiayu Wang, Hancheng Cao, and James Zou. 2025. The widespread adoption of large language model-assisted writing across society. arXiv preprint arXiv:2502.09747 (2025). [95] Xun Liang, Simin Niu, Zhiyu Li, Sensen Zhang, Hanyu Wang, Feiyu Xiong, Zhaoxin Fan, Bo Tang, Jihao Zhao, Jiawei Yang, et al. 2025. Saferag: Benchmarking security in retrieval-augmented generation of large language model. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 4609–4631. [96] Sheng-Chieh Lin, Akari Asai, Minghan Li, Barlas Oguz, Jimmy Lin, Yashar Mehdad, Wen-tau Yih, and Xilun Chen. 2023. How to train your dragon: Diverse augmentation towards generalizable dense retrieval. In Findings of the Association for Computational Linguistics: EMNLP 2023. 6385–6400. [97] Xi Victoria Lin, Xilun Chen, Mingda Chen, Weijia Shi, Maria Lomeli, Rich James, Pedro Rodriguez, Jacob Kahn, Gergely Szilvasy, Mike Lewis, et al. 2023. Ra-dit: Retrieval-augmented dual instruction tuning. arXiv preprint arXiv:2310.01352. [98] Fan Liu, Xiaozhuan Gao, Jie Zhao, and Yong Deng. 2019. Generalized belief entropy and its application in identifying conflict evidence. IEEe Access 7 (2019), 126625–126633. [99] Jingwei Liu, Ling Yang, Hongyan Li, and Shenda Hong. 2024. RetrievalAugmented Diffusion Models for Time Series Forecasting. In The Thirtyeighth Annual Conference on Neural Information Processing Systems. https: //openreview.net/forum?id=dRJJt0Ji48 [100] Nelson F Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. 2024. Lost in the middle: How language models use long contexts. Transactions of the association for computational linguistics 12 (2024), 157–173. [101] Tianyang Liu, Canwen Xu, and Julian J. McAuley. 2024. RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net. https://openreview.net/forum?id=pPjZIOuQuF [102] Yi Liu, Gelei Deng, Yuekang Li, Kailong Wang, Zihao Wang, Xiaofeng Wang, Tianwei Zhang, Yepang Liu, Haoyu Wang, Yan Zheng, et al. 2023. Prompt injection attack against llm-integrated applications. arXiv preprint arXiv:2306.05499 (2023). [103] Yupei Liu, Yuqi Jia, Runpeng Geng, Jinyuan Jia, and Neil Zhenqiang Gong. 2024. Formalizing and benchmarking prompt injection attacks and defenses. In 33rd USENIX Security Symposium (USENIX Security 24). 1831–1847. [104] Shuo Lu, Yinuo Xu, Jianjie Cheng, Lingxiao He, Meng Wang, and Jian Liang. 2026. DeepResearch-Slice: Bridging the Retrieval-Utilization Gap via Explicit Text Slicing. arXiv preprint arXiv:2601.03261 (2026). [105] Yongdong Luo, Xiawu Zheng, Guilin Li, Shukang Yin, Haojia Lin, Chaoyou Fu, Jinfa Huang, Jiayi Ji, Fei Chao, Jiebo Luo, and Rongrong Ji. 2025. VideoRAG: Visually-aligned Retrieval-Augmented Long Video Comprehension. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=QaZxGWlbgO
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
[106] Xinbei Ma, Yeyun Gong, Pengcheng He, Hai Zhao, and Nan Duan. 2023. Query rewriting in retrieval-augmented large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 5303– 5315. [107] Xueguang Ma, Xinyu Crystina Zhang, Ronak Pradeep, and Jimmy J. Lin. 2023. Zero-Shot Listwise Document Reranking with a Large Language Model. arXiv.org abs/2305.02156 (2023). [108] Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. 2024. Evaluating Very Long-Term Conversational Memory of LLM Agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Association for Computational Linguistics, 13851–13870. doi:10.18653/v1/2024.acl-long.747 [109] Alex Mallen, Akari Asai, Victor Zhong, Rajarshi Das, Daniel Khashabi, and Hannaneh Hajishirzi. 2023. When not to trust language models: Investigating effectiveness of parametric and non-parametric memories. In Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: Long papers). 9802–9822. [110] Kelong Mao, Zhicheng Dou, Fengran Mo, Jiewen Hou, Haonan Chen, and Hongjin Qian. 2023. Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search.. In Conference on Empirical Methods in Natural Language Processing (EMNLP). 1211–1225. [111] Shengyu Mao, Yong Jiang, Boli Chen, Xiao Li, Peng Wang, Xinyu Wang, Pengjun Xie, Fei Huang, Huajun Chen, and Ningyu Zhang. 2024. RaFe: Ranking Feedback Improves Query Rewriting for RAG. In Findings of the Association for Computational Linguistics: EMNLP 2024. 884–901. [112] Chuan Meng, Negar Arabzadeh, Mohammad Aliannejadi, and Maarten de Rijke. 2023. Query Performance Prediction: From Ad-hoc to Conversational Search. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. ACM, 2583–2593. [113] Jacob Menick, Maja Trebacz, Vladimir Mikulik, John Aslanides, Francis Song, Martin Chadwick, Mia Glaese, Susannah Young, Lucy Campbell-Gillingham, Geoffrey Irving, et al. 2022. Teaching language models to support answers with verified quotes. arXiv preprint arXiv:2203.11147 (2022). [114] Mike A Merrill, Alexander G Shaw, Nicholas Carlini, Boxuan Li, Harsh Raj, Ivan Bercovich, Lin Shi, Jeong Yeon Shin, Thomas Walshe, E Kelly Buchanan, et al. 2026. Terminal-bench: Benchmarking agents on hard, realistic tasks in command line interfaces. arXiv preprint arXiv:2601.11868 (2026). [115] Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Wei Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2023. FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation.. In Conference on Empirical Methods in Natural Language Processing (EMNLP). 12076–12100. [116] Yifei Ming, Senthil Purushwalkam, Shrey Pandit, Zixuan Ke, Xuan-Phi Nguyen, Caiming Xiong, and Shafiq Joty. [n. d.]. FaithEval: Can Your Language Model Stay Faithful to Context, Even If" The Moon is Made of Marshmallows". In The Thirteenth International Conference on Learning Representations. [117] Eric Mitchell, Yoonho Lee, Alexander Khazatsky, Christopher D Manning, and Chelsea Finn. 2023. Detectgpt: Zero-shot machine-generated text detection using probability curvature. In International conference on machine learning. PMLR, 24950–24962. [118] Fengran Mo, Zhan Su, Yuchen Hui, Jinghan Zhang, Jia Ao Sun, Zheyuan Liu, Chao Zhang, Tetsuya Sakai, and Jian-Yun Nie. 2026. OpenDecoder: Open Large Language Model Decoding to Incorporate Document Quality in RAG. CoRR abs/2601.09028 (2026). arXiv:2601.09028 doi:10.48550/arXiv.2601.09028 [119] Jesse Mu, Xiang Li, and Noah Goodman. 2023. Learning to compress prompts with gist tokens. Advances in Neural Information Processing Systems 36, 19327– 19352. [120] Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. 2021. Webgpt: Browser-assisted question-answering with human feedback. arXiv preprint arXiv:2112.09332 (2021). [121] Kanghui Ning, Zijie Pan, Yu Liu, Yushan Jiang, James Y. Zhang, Kashif Rasul, Anderson Schneider, Lintao Ma, Yuriy Nevmyvaka, and Dongjin Song. 2025. TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id= PymOnHw4Ty [122] Cheng Niu, Yuanhao Wu, Juno Zhu, Siliang Xu, Kashun Shum, Randy Zhong, Juntong Song, and Tong Zhang. 2024. Ragtruth: A hallucination corpus for developing trustworthy retrieval-augmented language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 10862–10878. [123] Rodrigo Nogueira and Kyunghyun Cho. 2019. Passage Re-ranking with BERT. arXiv.org abs/1901.04085 (2019). [124] Rodrigo Nogueira, Zhiying Jiang, Ronak Pradeep, and Jimmy Lin. 2020. Document ranking with a pretrained sequence-to-sequence model. In Findings of the association for computational linguistics: EMNLP 2020. 708–718.
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
[125] Kai Tzu-iunn Ong, Namyoung Kim, Minju Gwak, Hyungjoo Chae, Taeyoon Kwon, Yohan Jo, Seung-won Hwang, Dongha Lee, and Jinyoung Yeo. 2025. Towards lifelong dialogue agents via timeline-based memory management. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 8631–8661. [126] OpenAI. 2024. Introducing SWE-bench Verified. https://openai.com/index/ introducing-swe-bench-verified/. Accessed: 2026-02-12. [127] Jie Ouyang, Tingyue Pan, Mingyue Cheng, Ruiran Yan, Yucong Luo, Jiaying Lin, and Qi Liu. 2025. Hoh: A dynamic benchmark for evaluating the impact of outdated information on retrieval-augmented generation. arXiv preprint arXiv:2503.04800 (2025). [128] Charles Packer, Vivian Fang, Shishir_G Patil, Kevin Lin, Sarah Wooders, and Joseph_E Gonzalez. 2023. MemGPT: Towards LLMs as Operating Systems. (2023). [129] Liangming Pan, Wenhu Chen, Min-Yen Kan, and William Yang Wang. 2023. Attacking open-domain question answering by injecting misinformation. In Proceedings of the 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). 525–539. [130] Zhuoshi Pan, Qianhui Wu, Huiqiang Jiang, Menglin Xia, Xufang Luo, Jue Zhang, Qingwei Lin, Victor Rühle, Yuqing Yang, Chin-Yew Lin, et al. 2024. Llmlingua-2: Data distillation for efficient and faithful task-agnostic prompt compression. In Findings of the Association for Computational Linguistics: ACL 2024. 963–981. [131] Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. 2023. Generative Agents: Interactive Simulacra of Human Behavior. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology. ACM, 1–22. doi:10.1145/3586183.3606763 [132] Guilherme Penedo, Hynek Kydlíček, Anton Lozhkov, Margaret Mitchell, Colin A Raffel, Leandro Von Werra, Thomas Wolf, et al. 2024. The fineweb datasets: Decanting the web for the finest text data at scale. Advances in Neural Information Processing Systems 37 (2024), 30811–30849. [133] Guilherme Penedo, Quentin Malartic, Daniel Hesslow, Ruxandra Cojocaru, Hamza Alobeidli, Alessandro Cappelli, Baptiste Pannier, Ebtesam Almazrouei, and Julien Launay. 2023. The refinedweb dataset for falcon llm: Outperforming curated corpora with web data only. Advances in Neural Information Processing Systems 36 (2023), 79155–79172. [134] Wenjun Peng, Guiyang Li, Yue Jiang, Zilong Wang, Dan Ou, Xiaoyi Zeng, Derong Xu, Tong Xu, and Enhong Chen. 2024. Large Language Model based Long-tail Query Rewriting in Taobao Search. In Companion Proceedings of the ACM Web Conference 2024. ACM, 20–28. [135] Gustavo Penha and Claudia Hauff. 2021. On the Calibration and Uncertainty of Neural Learning to Rank Models for Conversational Search.. In Conference of the European Chapter of the Association for Computational Linguistics (EACL). 160–170. [136] A. Peysakhovich and Adam Lerer. 2023. Attention Sorting Combats Recency Bias In Long Context Language Models. arXiv.org abs/2310.01427 (2023). [137] Ofir Press, Muru Zhang, Sewon Min, Ludwig Schmidt, Noah A Smith, and Mike Lewis. 2023. Measuring and narrowing the compositionality gap in language models. In Findings of the Association for Computational Linguistics: EMNLP 2023. 5687–5711. [138] Zhen Qin, Rolf Jagerman, Kai Hui, Honglei Zhuang, Junru Wu, Le Yan, Jiaming Shen, Tianqi Liu, Jialu Liu, Donald Metzler, et al. 2024. Large language models are effective text rankers with pairwise ranking prompting. In Findings of the Association for Computational Linguistics: NAACL 2024. 1504–1518. [139] Xiangfei Qiu, Jilin Hu, Lekui Zhou, Xingjian Wu, Junyang Du, Buang Zhang, Chenjuan Guo, Aoying Zhou, Christian S Jensen, Zhenli Sheng, et al. 2024. TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods. Proceedings of the VLDB Endowment 17, 9 (2024), 2363–2377. [140] Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxiang Dong, Hua Wu, and Haifeng Wang. 2021. RocketQA: An optimized training approach to dense passage retrieval for open-domain question answering. In Proceedings of the 2021 conference of the North American chapter of the association for computational linguistics: human language technologies. 5835–5847. [141] Mohit Raghavendra, Anisha Gunjal, Bing Liu, and Yunzhong He. 2026. Agentic Rubrics as Contextual Verifiers for SWE Agents. arXiv preprint arXiv:2601.04171 (2026). [142] Revanth Gangi Reddy, JaeHyeok Doo, Yifei Xu, Md Arafat Sultan, Deevya Swain, Avirup Sil, and Heng Ji. 2024. FIRST: Faster improved listwise reranking with single token decoding. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 8642–8652. [143] Stephen Robertson, Hugo Zaragoza, et al. 2009. The probabilistic relevance framework: BM25 and beyond. Foundations and trends® in information retrieval 3, 4 (2009), 333–389. [144] Dongyu Ru, Lin Qiu, Xiangkun Hu, Tianhang Zhang, Peng Shi, Shuaichen Chang, Cheng Jiayang, Cunxiang Wang, Shichao Sun, Huanyu Li, et al. 2024. Ragchecker: A fine-grained framework for diagnosing retrieval-augmented
Lu Dai et al.
generation. Advances in Neural Information Processing Systems 37 (2024), 21999– 22027. [145] Alireza Salemi and Hamed Zamani. 2024. Evaluating retrieval quality in retrievalaugmented generation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2395–2400. [146] Keshav Santhanam, Omar Khattab, Jon Saad-Falcon, Christopher Potts, and Matei Zaharia. 2022. Colbertv2: Effective and efficient retrieval via lightweight late interaction. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 3715–3734. [147] Ahmad Sarlak and Abolfazl Razi. 2026. Reliability-Aware Determinantal Point Processes for Robust Informative Data Selection in Large Language Models. arXiv (2026). [148] Parth Sarthi, Salman Abdullah, Aditi Tuli, Shubh Khanna, Anna Goldie, and Christopher D Manning. 2024. Raptor: Recursive abstractive processing for tree-organized retrieval. In The Twelfth International Conference on Learning Representations. [149] Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselves to use tools. Advances in Neural Information Processing Systems 36 (2023), 68539–68551. [150] Avital Shafran, Roei Schuster, and Vitaly Shmatikov. 2025. Machine Against the { RAG } : Jamming { Retrieval-Augmented } Generation with Blocker Documents. In 34th USENIX Security Symposium (USENIX Security 25). 3787–3806. [151] Freda Shi, Xinyun Chen, Kanishka Misra, Nathan Scales, David Dohan, Ed H. Chi, Nathanael Schärli, and Denny Zhou. 2023. Large Language Models Can Be Easily Distracted by Irrelevant Context.. In International Conference on Machine Learning (ICML). 31210–31227. [152] Xiaoming Shi, Shiyu Wang, Yuqi Nie, Dianqi Li, Zhou Ye, Qingsong Wen, and Ming Jin. 2025. Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts. In The Thirteenth International Conference on Learning Representations. https://openreview.net/forum?id=e1wDDFmlVu [153] Zhengliang Shi, Yiqun Chen, Haitao Li, Weiwei Sun, Shiyu Ni, Yougang Lyu, Run-Ze Fan, Bowen Jin, Yixuan Weng, Minjun Zhu, et al. 2025. Deep research: A systematic survey. arXiv preprint arXiv:2512.02038 (2025). [154] Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. Advances in Neural Information Processing Systems 36 (2023), 8634–8652. [155] Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson, and Yarin Gal. 2024. AI models collapse when trained on recursively generated data. Nature 631, 8022 (2024), 755–759. [156] Michael D Skarlinski, Sam Cox, Jon M Laurent, James D Braza, Michaela Hinks, Michael J Hammerling, Manvitha Ponnapati, Samuel G Rodriques, and Andrew D White. 2024. Language agents achieve superhuman synthesis of scientific knowledge. arXiv preprint arXiv:2409.13740 (2024). [157] Luca Soldaini, Rodney Kinney, Akshita Bhagia, Dustin Schwenk, David Atkinson, Russell Authur, Ben Bogin, Khyathi Chandu, Jennifer Dumas, Yanai Elazar, et al. 2024. Dolma: An open corpus of three trillion tokens for language model pretraining research. In Proceedings of the 62nd annual meeting of the association for computational linguistics (volume 1: long papers). 15725–15788. MemoryGraft: Per[158] Saksham Sahai Srivastava and Haoyu He. 2025. sistent Compromise of LLM Agents via Poisoned Experience Retrieval. arXiv:2512.16962 [cs.CR] [159] Hongjin Su, Weijia Shi, Jungo Kasai, Yizhong Wang, Yushi Hu, Mari Ostendorf, Wen-tau Yih, Noah A Smith, Luke Zettlemoyer, and Tao Yu. 2023. One embedder, any task: Instruction-finetuned text embeddings. In Findings of the Association for Computational Linguistics: ACL 2023. 1102–1121. [160] Qisheng Su, Shiting Huang, Zhen Fang, Ziyan Chen, Zehui Chen, and Feng Zhao. 2026. Beyond Accuracy: Unveiling Inefficiency Patterns in Tool-Integrated Reasoning. arXiv preprint arXiv:2604.05404 (2026). [161] Weihang Su, Yichen Tang, Qingyao Ai, Zhijing Wu, and Yiqun Liu. 2024. DRAGIN: Dynamic Retrieval Augmented Generation based on the Real-time Information Needs of Large Language Models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 12991–13013. [162] Weiwei Sun, Lingyong Yan, Xinyu Ma, Shuaiqiang Wang, Pengjie Ren, Zhumin Chen, Dawei Yin, and Zhaochun Ren. 2023. Is ChatGPT Good at Search? Investigating Large Language Models as Re-Ranking Agents.. In Conference on Empirical Methods in Natural Language Processing (EMNLP). 14918–14937. [163] Liyan Tang, Philippe Laban, and Greg Durrett. 2024. Minicheck: Efficient factchecking of llms on grounding documents. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 8818–8847. [164] Nandan Thakur, Nils Reimers, Andreas Rücklé, Abhishek Srivastava, and Iryna Gurevych. [n. d.]. BEIR: A Heterogeneous Benchmark for Zero-shot Evaluation of Information Retrieval Models. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2). [165] Suxin Tong and Jingling Yuan. 2025. Efficiently Enhancing Long-term Series Forecasting via Ultra-long Lookback Windows. In Proceedings of the AAAI
LLM-Oriented Information Retrieval: A Denoising-First Perspective
Conference on Artificial Intelligence, Vol. 39. 20912–20920. [166] Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2023. Interleaving retrieval with chain-of-thought reasoning for knowledgeintensive multi-step questions. In Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers). 10014–10037. [167] Yiteng Tu, Weihang Su, Yujia Zhou, Yiqun Liu, and Qingyao Ai. 2025. Robust Fine-tuning for Retrieval Augmented Generation against Retrieval Defects. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2025, Padua, Italy, July 13–18, 2025. ACM, 1272–1282. doi:10.1145/3726302.3730078 [168] Kirill Vishniakov, Boulbaba Ben Amor, Engin Tekin, Nancy A. ElNaker, Karthik Viswanathan, Aleksandr Medvedev, Aahan Singh, Maryam Nadeem, Mohammad Amaan Sayeed, Praveenkumar Kanithi, Tiago Magalhaes, Natalia Vassilieva, Dwarikanath Mahapatra, Marco AF Pimentel, and Shadab Khan. 2025. Gene42: Long-Range Genomic Foundation Model With Dense Attention. In NeurIPS 2025 Workshop on AI Virtual Cells and Instruments: A New Era in Drug Discovery and Development. https://openreview.net/forum?id=xztaGxabHV [169] Fei Wang, Xingchen Wan, Ruoxi Sun, Jiefeng Chen, and Sercan O Arik. 2025. Astute rag: Overcoming imperfect retrieval augmentation and knowledge conflicts for large language models. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 30553–30571. [170] Han Wang, Archiki Prasad, Elias Stengel-Eskin, and Mohit Bansal. 2025. Retrieval-augmented generation with conflicting evidence. arXiv preprint arXiv:2504.13079 (2025). [171] Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533 (2022). [172] Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2023. Simlm: Pre-training with representation bottleneck for dense passage retrieval. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2244– 2258. [173] Liang Wang, Nan Yang, and Furu Wei. 2023. Query2doc: Query Expansion with Large Language Models.. In Conference on Empirical Methods in Natural Language Processing (EMNLP). 9414–9423. [174] Xiaohua Wang, Zhenghua Wang, Xuan Gao, Feiran Zhang, Yixin Wu, Zhibo Xu, Tianyuan Shi, Zhengyuan Wang, Shizheng Li, Qi Qian, et al. 2024. Searching for best practices in retrieval-augmented generation. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 17716–17736. [175] You Wang, Michael Pradel, and Zhongxin Liu. 2025. Are “Solved Issues” in SWEbench Really Solved Correctly? An Empirical Study. arXiv:2503.15223 [cs.SE] http://arxiv.org/abs/2503.15223v2 [176] Zhiruo Wang, Jun Araki, Zhengbao Jiang, Md Rizwan Parvez, and Graham Neubig. 2023. Learning to filter context for retrieval-augmented generation. arXiv preprint arXiv:2311.08377 (2023). [177] Zifeng Wang, Lang Cao, Qiao Jin, Joey Chan, Nicholas Wan, Behdad Afzali, Hyun-Jin Cho, Chang-In Choi, Mehdi Emamverdi, Manjot K. Gill, Sun-Hyung Kim, Yijia Li, Yi Liu, Yiming Luo, Hanley Ong, Justin F. Rousseau, Irfan Sheikh, Jenny J. Wei, Ziyang Xu, Christopher M. Zallek, Kyungsang Kim, Yifan Peng, Zhiyong Lu, and Jimeng Sun. 2025. A foundation model for human-AI collaboration in medical literature mining. Nature Communications 16, 1 (2025). [178] Zhitong Wang, Cheng Gao, Chaojun Xiao, Yufei Huang, Shuzheng Si, Kangyang Luo, Yuzhuo Bai, Wenhao Li, Tangjian Duan, Chuancheng Lv, et al. 2025. Document segmentation matters for retrieval-augmented generation. In Findings of the Association for Computational Linguistics: ACL 2025. 8063–8075. [179] Zilong Wang, Zifeng Wang, Long Le, Steven Zheng, Swaroop Mishra, Vincent Perot, Yuwei Zhang, Anush Mattapalli, Ankur Taly, Jingbo Shang, Chen-Yu Lee, and Tomas Pfister. 2025. Speculative RAG: Enhancing Retrieval Augmented Generation through Drafting. In The Thirteenth International Conference on Learning Representations, Vol. abs/2407.08223. [180] Ziyang Wang, Shoubin Yu, Elias Stengel-Eskin, Jaehong Yoon, Feng Cheng, Gedas Bertasius, and Mohit Bansal. 2025. VideoTree: Adaptive Tree-based Video Representation for LLM Reasoning on Long Videos. In Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR). 3272–3283. [181] Orion Weller, Michael Boratko, Iftekhar Naim, and Jinhyuk Lee. 2025. On the theoretical limitations of embedding-based retrieval. arXiv preprint arXiv:2508.21038 (2025). [182] Orion Weller, Aleem Khan, Nathaniel Weir, Dawn Lawrie, and Benjamin Van Durme. 2024. Defending against disinformation attacks in open-domain question answering. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Papers). 402–417. [183] Guillaume Wenzek, Marie-Anne Lachaux, Alexis Conneau, Vishrav Chaudhary, Francisco Guzmán, Armand Joulin, and Edouard Grave. 2020. CCNet: Extracting high quality monolingual datasets from web crawl data. In Proceedings of the twelfth language resources and evaluation conference. 4003–4012. [184] Di Wu, Hongwei Wang, Wenhao Yu, Yuwei Zhang, Kai-Wei Chang, and Dong Yu. 2025. LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
Memory. In ICLR. arXiv:2410.10813 [cs.CL] [185] Chunqiu Steven Xia, Yinlin Deng, Soren Dunn, and Lingming Zhang. 2025. Demystifying LLM-Based Software Engineering Agents. Proceedings of the ACM on Software Engineering 2, FSE (2025), 801–824. doi:10.1145/3715754 [186] Chong Xiang, Tong Wu, Zexuan Zhong, David Wagner, Danqi Chen, and Prateek Mittal. 2024. Certifiably robust rag against retrieval corruption. arXiv preprint arXiv:2405.15556 (2024). [187] Jian Xie, Kai Zhang, Jiangjie Chen, Renze Lou, and Yu Su. 2023. Adaptive chameleon or stubborn sloth: Revealing the behavior of large language models in knowledge conflicts. In The Twelfth International Conference on Learning Representations. [188] Lee Xiong, Chenyan Xiong, Ye Li, Kwok-Fung Tang, Jialin Liu, Paul N Bennett, Junaid Ahmed, and Arnold Overwijk. [n. d.]. Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval. In International Conference on Learning Representations. [189] Fangyuan Xu, Weijia Shi, and Eunsol Choi. 2024. RECOMP: Improving retrievalaugmented LMs with context compression and selective augmentation. In The Twelfth International Conference on Learning Representations. [190] Shicheng Xu, Liang Pang, Huawei Shen, and Xueqi Cheng. 2025. A Theory for Token-Level Harmonization in Retrieval-Augmented Generation. In The Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025. OpenReview.net. https://openreview.net/forum?id= tbx3u2oZAu [191] John Yang, Carlos E. Jimenez, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik Narasimhan, and Ofir Press. 2024. SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering. In Advances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024. http://papers.nips.cc/paper_files/paper/2024/hash/ 5a7c947568c1b1328ccc5230172e1e7c-Abstract-Conference.html [192] Zuhao Yang, Yingchen Yu, Yunqing Zhao, Shijian Lu, and Song Bai. 2025. Timeexpert: An expert-guided video llm for video temporal grounding. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 24286–24296. [193] Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. In The eleventh international conference on learning representations. [194] Fanghua Ye, Meng Fang, Shenghui Li, and Emine Yilmaz. 2023. Enhancing Conversational Search: Large Language Model-Aided Informative Query Rewriting.. In Conference on Empirical Methods in Natural Language Processing (EMNLP). 5985–6006. [195] Ori Yoran, Tomer Wolfson, Ori Ram, and Jonathan Berant. [n. d.]. Making Retrieval-Augmented Language Models Robust to Irrelevant Context. In The Twelfth International Conference on Learning Representations. [196] Boxi Yu, Yuxuan Zhu, Pinjia He, and Daniel Kang. 2025. UTBoost: Rigorous Evaluation of Coding Agents on SWE-Bench. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2025, Vienna, Austria, July 27 - August 1, 2025. Association for Computational Linguistics, 3762–3774. https://aclanthology.org/2025.acl-long.189/ [197] Tan Yu, Anbang Xu, and Rama Akkiraju. 2024. In defense of rag in the era of long-context language models. arXiv preprint arXiv:2409.01666 (2024). [198] Wenhao Yu, Dan Iter, Shuohang Wang, Yichong Xu, Mingxuan Ju, Soumya Sanyal, Chenguang Zhu, Michael Zeng, and Meng Jiang. [n. d.]. Generate rather than Retrieve: Large Language Models are Strong Context Generators. In The Eleventh International Conference on Learning Representations. [199] W. Yu, Hongming Zhang, Xiaoman Pan, Kaixin Ma, Hongwei Wang, and Dong Yu. 2023. Chain-of-Note: Enhancing Robustness in Retrieval-Augmented Language Models. In Conference on Empirical Methods in Natural Language Processing, Vol. abs/2311.09210. [200] Yi Yu, Liuyi Yao, Yuexiang Xie, Qingquan Tan, Jiaqi Feng, Yaliang Li, and Libing Wu. 2026. Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents. arXiv:2601.01885 [cs.CL] [201] Yuheng Zha, Yichi Yang, Ruichen Li, and Zhiting Hu. 2023. AlignScore: Evaluating factual consistency with a unified alignment function. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 11328–11348. [202] ChengXiang Zhai. 2025. Information Retrieval for Artificial General Intelligence: A New Perspective of Information Retrieval Research. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 3876–3886. [203] Bowen Zhang and Harold Soh. 2024. Extract, Define, Canonicalize: An LLMbased Framework for Knowledge Graph Construction. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 9820–9836. [204] Fengji Zhang, Bei Chen, Yue Zhang, Jacky Keung, Jin Liu, Daoguang Zan, Yi Mao, Jian-Guang Lou, and Weizhu Chen. 2023. RepoCoder: Repository-Level Code Completion Through Iterative Retrieval and Generation. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics, 2471–2484. doi:10.18653/v1/2023.
SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia
emnlp-main.151 [205] Fangyuan Zhang, Zhengjun Huang, Yingli Zhou, Qintian Guo, Zhixun Li, Wensheng Luo, Di Jiang, Yixiang Fang, and Xiaofang Zhou. 2025. EraRAG: Efficient and Incremental Retrieval Augmented Generation for Growing Corpora. arXiv preprint arXiv:2506.20963 (2025). [206] Guangwei Zhang, Qisheng Su, Jiateng Liu, Cheng Qian, Yanzhou Pan, Yanjie Fu, and Denghui Zhang. 2025. ISACL: Internal State Analyzer for Copyrighted Training Data Leakage. In Findings of the Association for Computational Linguistics: EMNLP 2025. [207] Guangwei Zhang, Jianing Zhu, Cheng Qian, Neil Gong, Rada Mihalcea, Zhaozhuo Xu, Jingrui He, Jiaqi Ma, Yun Huang, Chaowei Xiao, et al. 2026. Copyright Detective: A Forensic System to Evidence LLMs Flickering Copyright Leakage Risks. arXiv preprint arXiv:2602.05252 (2026). [208] Hang Zhang, Yeyun Gong, Yelong Shen, Jiancheng Lv, Nan Duan, and Weizhu Chen. [n. d.]. Adversarial Retriever-Ranker for Dense Text Retrieval. In International Conference on Learning Representations. [209] Hairong Zhang, Jiaheng Si, Guohang Yan, Boyuan Qi, Pinlong Cai, Song Mao, Ding Wang, and Botian Shi. 2025. RAKG: Document-level Retrieval Augmented Knowledge Graph Construction. arXiv preprint arXiv:2504.09823 (2025). [210] Kaichen Zhang, Yuzhong Hong, Junwei Bao, Hongfei Jiang, Yang Song, Dingqian Hong, and Hui Xiong. 2025. Gvpo: Group variance policy optimization for large language model post-training. arXiv preprint arXiv:2504.19599 (2025). [211] Mingtian Zhang, Yu Tang, and PageIndex Team. 2025. PageIndex: NextGeneration Vectorless, Reasoning-based RAG. PageIndex Blog (September 2025). https://pageindex.ai/blog/pageindex-intro. [212] Ming Zhang, Jiabao Zhuang, Wenqing Jing, Ziyu Kong, Jingyi Deng, Yujiong Shen, Kexin Tan, Yuhang Zhao, Ning Luo, Renzhe Zheng, et al. 2026. Can Deep Research Agents Find and Organize? Evaluating the Synthesis Gap with Expert Taxonomies. arXiv preprint arXiv:2601.12369 (2026). [213] Quan Zhang, Binqi Zeng, Chijin Zhou, Gwihwan Go, Heyuan Shi, and Yu Jiang. 2024. Human-imperceptible retrieval poisoning attacks in LLM-powered applications. In Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering. 502–506. [214] Shengtao Zhang, Jiaqian Wang, Ruiwen Zhou, Junwei Liao, Yuchen Feng, Weinan Zhang, Ying Wen, Zhiyu Li, Feiyu Xiong, Yutao Qi, et al. 2026. MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory. arXiv preprint arXiv:2601.03192 (2026). [215] Zhuocheng Zhang, Yang Feng, and Min Zhang. 2025. LevelRAG: Enhancing Retrieval-Augmented Generation with Multi-hop Logic Planning over Rewriting
Lu Dai et al.
Augmented Searchers. arXiv preprint arXiv:2502.18139 (2025). [216] Huaixiu Steven Zheng, Swaroop Mishra, Xinyun Chen, Heng-Tze Cheng, Ed H Chi, Quoc V Le, and Denny Zhou. [n. d.]. Take a Step Back: Evoking Reasoning via Abstraction in Large Language Models. In The Twelfth International Conference on Learning Representations. [217] Yuxiang Zheng, Dayuan Fu, Xiangkun Hu, Xiaojie Cai, Lyumanshan Ye, Pengrui Lu, and Pengfei Liu. 2025. Deepresearcher: Scaling deep research via reinforcement learning in real-world environments. arXiv preprint arXiv:2504.03160 (2025). [218] Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. 2024. MemoryBank: Enhancing Large Language Models with Long-Term Memory. Proceedings of the AAAI Conference on Artificial Intelligence 38, 17 (2024), 19724–19731. doi:10.1609/aaai.v38i17.29946 [219] Jiawei Zhou and Lei Chen. 2025. OpenRAG: Optimizing RAG End-to-End via In-Context Retrieval Learning. arXiv preprint arXiv:2503.08398 (2025). [220] Luowei Zhou, Chenliang Xu, and Jason Corso. 2018. Towards automatic learning of procedures from web instructional videos. In Proceedings of the AAAI conference on artificial intelligence, Vol. 32. [221] Rongzhi Zhu, Xiangyu Liu, Zequn Sun, Yiwei Wang, and Wei Hu. 2025. Mitigating lost-in-retrieval problems in retrieval augmented multi-hop question answering. arXiv preprint arXiv:2502.14245 (2025). [222] Xiangrong Zhu, Yuexiang Xie, Yi Liu, Yaliang Li, and Wei Hu. 2025. Knowledge Graph-Guided Retrieval Augmented Generation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 8912–8924. [223] Zulun Zhu, Haoyu Liu, Mengke He, and Siqiang Luo. 2025. Right Answer at the Right Time-Temporal Retrieval-Augmented Generation via Graph Summarization. arXiv preprint arXiv:2510.16715 (2025). [224] Shengyao Zhuang, Honglei Zhuang, Bevan Koopman, and Guido Zuccon. 2024. A setwise approach for effective and highly efficient zero-shot ranking with large language models. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 38–47. [225] Wei Zou, Runpeng Geng, Binghui Wang, and Jinyuan Jia. 2025. { PoisonedRAG } : Knowledge corruption attacks to { Retrieval-Augmented } generation of large language models. In 34th USENIX Security Symposium (USENIX Security 25). 3827–3844.