Retrieval Augmented Generation (RAG) is a key component for generating accurate and hallucination free answers using Large Language Models (LLMs). LLMs are improving at handling long context, but still suffer from "lost in the middle" problem. Thus, precise and accurate retrieval is important. Current retrievers chunk long context into length-based manageable chunks - in the process throwing away rich and informative semantic global structure in the corpus. We introduce a novel retrieval system STAIR that empowers an LLM to exploit global structure in a corpus such as a Table of Contents (ToC) to efficiently store and retrieve information from its model parameters. Our thorough and careful ablation studies with a finetuned Differentiable Search Index (DSI) system show that ToC helps build a low hallucination (less than 0.05%) generative Information Retrieval (IR) system and can generalize to examples where very few training samples are available. To further research in this novel direction of ToC based retrieval we release SearchTome - a diverse benchmark created from 18 books across 6 diverse domains to further research in this novel direction. STAIR achieves a high Recall@1 score of 82.6% on SearchTome as compared to DSI (76.9%), where the difference is found to be statistically significant. STAIR easily beats other strong baselines such as BM25 (59.5%), DPR (68.7%) and out-of-the-box Mistral (13.8%).
Libraries and archives manage large collections with limited staff and computing budgets, yet common benchmarks do not systematically test their bibliographic work. They need to know which methods work for their tasks and what those methods require to run. SHELF, the Synthetic Harness for Evaluating LLM Fitness, addresses this gap. It is a Python system that turns labelled taxonomies, writing specifications, and a generation budget into controlled benchmark data and evaluation tasks. This first release contains 62,899 model-written documents based on Library of Congress vocabularies, with tasks for classification, clustering, retrieval, pair classification, and instruction retrieval. We compare TF, TF-IDF, BM25, popular encoders, and, on subject classification only, zero-shot decoders; each method appears only on tasks that support it. Subject classification reaches 0.8887, while genre-form classification reaches only 0.2605, and several pair and clustering tasks remain near chance. Sparse methods remain competitive on classification, while TF-IDF is the fastest measured arm in the subject timing experiment. SHELF also varies bibliographic facets independently and can generate new, verifiably unseen documents after a model's training cutoff. Comparisons with LCSHBench and Project Gutenberg show that model rankings transfer more reliably than absolute scores, but SHELF scores do not estimate accuracy on production catalogue data. We release all source code and data under permissive licenses on GitHub and Hugging Face.
Runpeng Dai, Kaili Huang, Changsung Kang +1cs.IR cs.CL
Retrieval is the first stage of modern search and advertising systems, selecting a candidate set from a large item universe for downstream ranking and auction. Recent work increasingly leverages LLMs to improve retrieval through query expansion, data synthesis, and retrieval-feedback training. However, the generative component is typically used for query-side augmentation, while final matching is still delegated to a downstream retriever. We introduce CoGR, a retrieval framework that instead trains LLMs to directly construct retrieval representations on both query and item sides. Each generator produces a compact set of keywords, which are matched directly through an inverted index, preserving compatibility with existing keyword-based retrieval infrastructure. CoGR uses a two-stage training pipeline. Supervised fine-tuning first establishes an aligned keyword space, after which co-evolving reinforcement learning alternately optimizes the query- and item-side generators with GRPO against the opposite side's frozen index. Both sides optimize the same query-to-item retrieval $F_1$ objective: the query side receives retrieval $F_1$ directly, while the item side receives a counterfactual marginal reward measuring the change in query-side $F_1$ caused by its generated keywords. Across 10 representative sparse, dense, and generative baselines, CoGR achieves the best performance on both an internal APP Marketplace dataset and the public WANDS benchmark, improving $F_1$ over the strongest baseline by $10.9\%$ and $36.1\%$, respectively. Further analysis shows stable co-evolution and increasingly aligned query--item keyword spaces over training.
Retrieval-augmented language models can fail to respect negative constraints when the retriever supplies evidence about concepts the user explicitly excluded. Beyond explicit negation, queries may ask for answers that include one concept while excluding another, or for entities that belong to a category but differ from a closely related instance. Because the excluded concept still appears in the query text, dense retrievers may assign high similarity to documents about that concept even when the user asks to avoid it. We introduce E-SENS, a training-free reranking method for negation-sensitive retrieval. E-SENS extracts a compact trap query for the excluded side and subtracts trap-query similarity from the original-query retrieval score. On ExcluIR, E-SENS shows a clear recall-violation trade-off across four embedding models and reduces trap retrieval at recall-preserving settings.
Retrieved scientific literature can serve as inspiration for both human and AI scientists. Inspiration can take different forms: prior work may directly suggest how to address a problem, or surface directions at different levels of abstraction - zooming out to a more general view or zooming in to a concrete realization. We introduce RATIO (Retrieval Across Typed Ideation Operations), a large-scale benchmark in which relevance is defined by three operations which we name ideation moves: Address retrieves potential approaches for stated problems, Broaden retrieves more general formulations, and Specify retrieves concrete instantiations. RATIO is constructed from millions of full-text scientific papers across CS literature via a general recipe that extends discourse-marker distant supervision - previously used only for classification - to corpus-scale retrieval, combined with extensive LLM and human vetting. Experiments show that operation-specific fine-tuning substantially boosts retrievers but leaves much room for further improvements. RATIO provides a scalable training and evaluation framework for retrieval components that support literature-grounded ideation, opening up new research avenues on scientific inspiration retrieval.
Theodore Rogers, Joe Standerfer, Dmitrii Timoshenko +3cs.IR cs.LG
A shared search-and-recommendation index must score new items from features alone because search has no exploration slot. In a public log covering both surfaces over one catalog, $38.6\%$ of held-out query-search impressions show an item never previously shown or visited. For user-cold engagements, the feature-based tower serves this demand without measurable loss against $99$ sampled negatives ($0.9595$ Recall@20 versus $0.9510$ warm). A lexical baseline reaches similar parity, while a full-catalog check remains statistically undecided. Dual-encoder retrieval therefore keeps the index \emph{open} to new items, unlike an ID-softmax recommender that requires retraining. We price this openness on recommendation against six sequential baselines, each retrained and tuned through five rounds on corrected targets. A float32 timestamp bug had reordered leave-one-out targets for $19.7\%$ of users. On MovieLens-1M, warm accuracy trails the strongest retrained baseline by $5.2\%$ Recall@20 and $11.4\%$ NDCG@20. On MIND, the gap narrows to $0.8$--$3.6\%$ relative to the five strongest baselines, though the model ranks sixth of seven. Under strict zero-leakage cold-start evaluation, the content tower achieves $0.172 \pm 0.006$ Recall@20, $1.4\times$ the strongest retrained dedicated method ($0.124 \pm 0.007$) and $3\times$ a training-free floor, without cold-specific training. Exact full-softmax training raises Recall@20 by $54\%$ on MIND-small and $6.9\%$ on MovieLens-1M over sampled InfoNCE, but recomputes the full catalog each step and exhausts accelerator memory at $240$K items. Approximate nearest-neighbor search explains none of the remaining gap, serving cost does not regress against ID-softmax retrieval, and a history-window sweep explains half the post-recipe remainder. Exact-quality training at catalog scale remains the open problem.
Historical retrieval for time-series prediction commonly treats past similarity as a proxy for usefulness. We ask a different question: which historical examples should be expected to matter for a query? We define predictive relevance as expected future utility conditioned on inference-time information, using realized futures only during training as privileged supervision. A normalized-pattern retriever first forms a coarse candidate set, and a lightweight residual multilayer perceptron (MLP) learns a listwise future-compatibility target while keeping inference-time scoring strictly past-only. Our method retains similarity-based candidate generation but reranks its candidates by a more predictive relevance criterion. Optimal relevance decomposes into candidate-level utility and query-specific compatibility, motivating Candidate-Prior and Shuffled-Future controls. Across six benchmarks, the reranker improves Pattern retrieval while revealing candidate-global, query-specific, and mixed relevance regimes. On all 12 confirmatory tasks, it improves Pattern and outperforms a matched-protocol Stationarity-Aware Retrieval-Augmented Time Series Forecasting (SARAF) retrieval rule. Architecture-matched ablations show that correct future supervision, rather than the MLP or added context alone, drives gains in query-specific regimes. Alternative-similarity experiments show that a strong last-value-anchored L2 rule remains superior in some domains, whereas future-supervised relevance is particularly strong where our diagnostics indicate query-specific relevance, especially on Solar. Candidate-pool diagnostics show that this contrast is not explained solely by coarse Pattern retrieval. Overall, historical relevance is structured and domain dependent rather than governed by a universally superior retrieval rule.
Enterprise repositories are large, heteroge- neous, and continuously updated, making re- trieval difficult when efficient access, source- faithful evidence, and cross-query adaptation must be supported together. Enterprise mem- ory extends retrieval beyond the model con- text, but existing systems do not jointly address collection-specific hierarchy construction, low- cost routing, detailed evidence loading, and workload-aware memory updates at this scale. We introduce MEMONDEMAND, short for On- Demand Memory, a memory management sys- tem with three coordinated mechanisms: a dy- namic multi-level hierarchy that determines the abstraction structure and depth for each col- lection, dual memory at every hierarchy level that separates distilled routing from detailed evidence, and on-demand memory promotion that updates node priority under a bounded active-state budget. On EnterpriseRAG-Bench, MEMONDEMAND outperforms the strongest published LB#1 result at every evaluated scale from 10M tokens through the complete 618M- token collection, with gains of 12.23% at 10M and 4.66% at 618M. Results on FinanceBench, HotpotQA, and FRAMES further show strong performance across financial, multi-hop, and fact-retrieval settings. Together, these results establish MEMONDEMAND as an accurate, ef- ficient, and scalable memory solution for very large enterprise repositories across data scales, domains, and evidence requirements. Our code is available at https://github.com/ xfab-xinyuansong/MemOnDemand.git.
Recent works have highlighted the expressive limitations of embedding based retrieval models through both theoretical analyses and challenging benchmarks such as LIMIT. While multi-vector embeddings consistently outperform single-vector embeddings, the precise representational gap between them remains poorly understood. In this work, following Jayaram's work, we provide the first explicit family of query and document sets, together with their relevance matrices, for which single-vector embeddings that rank all relevant documents above irrelevant ones require exponential size, whereas polynomial-size multi-vector embeddings suffice. Our result establishes an exponential separation between the expressive power of single-vector and multi-vector embeddings for the task of ranking of documents as opposed to approximating numerical scores as in the work of Jayaram. Motivated by our theoretical construction, we introduce ANDOR, a new retrieval benchmark that naturally instantiates these hard examples. We show that state-of-the-art single-vector embedding models perform poorly on ANDOR in the zero-shot setting and exhibit only marginal improvements after fine-tuning, highlighting the inherent difficulty of the benchmark compared to prior work. In contrast, multi-vector models consistently outperform their single-vector counterparts and improve substantially with fine-tuning, closely aligning with our theoretical predictions.
More context does not require a larger retrieval budget. Under the same ceiling, a retrieval system can recover more of the evidence a question requires by following relationships between evidence that flat top-k ranking leaves behind. We test that proposition with Dual-Bounded Relational Recall (DBRR), which allocates a fixed retrieval budget between relevance-selected seeds and bounded graph-adjacent context, against matched flat top-k retrieval using the same relevance-ranking stage and the same maximum number of retrieval units and tokens. The outcome is complete recovery of the official HotpotQA supporting-evidence set for each question. Across 7,405 FullWiki questions, the Primary DBRR allocation increased complete supporting-evidence recovery by 23.8 percentage points over its matched flat baseline (paired risk difference 0.2377; question-level bootstrap 95% interval 0.2269 to 0.2489). It improved 1,952 questions, tied on 5,261, and harmed 192. Bridge questions drove the effect, with a 28.7-point increase; comparison questions showed a smaller 4.2-point difference. In a prespecified, evaluation-only diagnostic population, real relationships also outperformed random-neighbor and degree-preserving shuffled-graph controls. The result is straightforward: under the same context budget, complete-evidence retrieval depends not only on which items rank highest, but on how context is allocated around them. Relational allocation recovered complete evidence sets that flat top-k retrieval left incomplete.
Modern LLMs excel at reasoning and instruction following, enabling users to express complex and diverse information needs. However, conventional retrievers largely rely on surface-level matching between queries and documents, resulting in a growing gap between how users express their needs and how retrievers interpret them. In this paper, we present GEM, a generative embedding model that augments retrieval through its own knowledge by explicitly reasoning about user intent and relevance criteria. GEM unifies generation and embedding within a single model: it first reasons over the query, then appends an embedding token to encode the enriched context for retrieval. \zhili{Evaluated on reasoning-intensive and instruction-following retrieval tasks, GEM demonstrates the effectiveness of its reasoning-augmented retrieval, outperforming its non-reasoning variant and matching baselines using substantially larger models.} Furthermore, GEM's generative nature allows test-time compute scaling via prompting to further enhance retrieval performance. Our code is available at: https://anonymous.4open.science/r/GEM.
Rose Cymbler, Daniel Guez, Laurent Fabrecs.CL cs.AI cs.IR
We identify and quantify temporal misgrounding: the systematic retrieval and citation of the currently in-force version of a legal article when the applicable version is an earlier or future one. Standard legal RAG treats the corpus as static; we argue legal question answering is a temporally-indexed retrieval problem. We introduce FiscalQA Pro, pairing a versioned corpus of 32,436 article-versions of the French tax code (93 years, 1938-2031) with an all-model-hard temporal-reasoning track: 209 scored, expert-reviewed questions across 33 CGI articles (221 released; twelve flagged out of the answerable scope). At selection time, no evaluated model recovered its date-applicable answer closed-book in any of four sampling draws, and the currently in-force text lacks the gold value for all but one of the scored questions. Answers are scored deterministically via atomic ground-truth "nuggets" (regex and numeric-with-tolerance), never LLM-as-judge: an LLM judge would inherit the temporal bias it is meant to score. Across eleven models (five frontier closed-API systems plus Gemini 2.5 Pro as a substitute entry, and five open-weight), parametric knowledge yields 3.0% mean strict accuracy and RAG over a static current-version corpus 2.7%. Static RAG retrieves the date-applicable version 0% of the time, confidently citing a real but inapplicable version. Our end-to-end retriever over a multi-version index, with no oracle, reaches 98.3% mean strict; an oracle-article ablation reaches 99.1%, locating the residual gap in first-stage recall, not version selection. We additionally release a version-aware jurisprudence dataset of 69,208 citation links, together with the corpus, benchmark, model responses, and pipeline code.
Scientific literature is increasingly used as a knowledge source for language models, retrieval-augmented generation systems, and research assistants, but answering research questions from papers requires more than fluent generation. A reliable system must identify the relevant papers, locate the concrete evidence that supports the answer, and produce a response that is faithful to that evidence. We present LitTraceQA, a benchmark for literature-grounded question answering over scientific papers. Given a research question and a metadata pool of papers, a system must return three connected outputs: canonical paper identifiers, supporting evidence locations, and answers in one or more requested formats, including free-form text, multiple-choice answers, and structured tables. LitTraceQA targets evidence types common in scientific reading: tables, figures, text spans, equations or algorithms, and citation contexts. The public development split contains 55 examples, including 26 hidden-source single-paper questions and 29 multi-paper questions, and provides gold papers, evidence annotations, and answers for local validation. We also analyze a larger final annotation collection with 4,978 unique-question records over 4,859 unique gold papers. By evaluating paper retrieval, evidence grounding, and answer accuracy separately, LitTraceQA provides a testbed for scientific QA systems that produce verifiable answers rather than unsupported summaries.
Linhai Ma, Ethan F. Wei, Xueqing Peng +3cs.CL cs.AI
Large-taxonomy retrieval often assumes that the input already expresses the target concept. In many settings, however, the input is indirect evidence, such as a table cell whose meaning depends on its row, column, datatype, and context. We call this mismatch the retrieval readiness gap. Our analysis shows that the current index retrieves the target reliably when its semantics are explicit, while raw evidence often leaves it deep in the ranking. We propose Factorized Hypothesis Search (FHS), which maintains multiple partial interpretations over named semantic dimensions. These hypotheses support structured query rendering, multi-hypothesis retrieval, and dimension-level candidate verification. On both financial taxonomy tagging and CodiEsp clinical coding tasks, FHS achieves the best Recall@1, MRR, and final accuracy among the non-oracle methods. Replacing the factorized hypothesis path with a free-text ensemble causes the largest drop in head-ranking performance, while sequential refinement provides no additional gain over FHS's strong parallel first round.
Trajectory similarity learning is fundamental to efficient trajectory retrieval under complex distance measures. Existing learning-based methods typically rely on embeddings trained to approximate trajectory distances or rankings, but they often lack guarantees with respect to the original distances, exhibit unstable performance across distance measures, and incur substantial training costs. We revisit trajectory similarity learning from a lower-bound representation perspective and propose LB-TrajRep, a unified lower-bound representation framework independent of deep neural embeddings. This framework constructs single-vector representations from a set of lower-bound components, enabling admissible and interpretable lower bounds for multiple classical trajectory distances, including Dynamic Time Warping (DTW), Hausdorff distance, and Discrete Fréchet Distance (DFD). Within this framework, we instantiate point-pivot components, which naturally support both metric and non-metric distances and remain compatible with standard vector-based retrieval pipelines. To improve ranking quality, we develop two data-driven pivot selection strategies that explicitly optimize lower-bound tightness and prioritize hard near-neighbor trajectory pairs, respectively. Extensive experiments on real-world trajectory datasets show that the proposed lower-bound representations are able to consistently outperform state-of-the-art neural trajectory embeddings across diverse distance measures, improving top-$k$ ranking accuracy by up to 20\%--60\% on the Hausdorff distance and DFD and by 15\%--40\% on DTW.
Enterprise data analytics agents face two structural failures: generic RAG retrieves the wrong asset (Hit@10=19.1%) and delivers no usage knowledge to prevent metric misinterpretation---stemming from four root causes (C1--C4) ranging from semantic gap and entity ambiguity to schema drift and asset-usage gap. We present a two-layer solution deployed in the commercial advertising data warehouse at Xiaohongshu (5,300+ Hive tables, 14 domains). A three-tier dual-purpose knowledge base (179 documents, eight-section annotation template) serves both retrieval and generation, with a closed-loop refresh pipeline maintaining day-level freshness (one yes/no approval, 30s hot-reload). The Graph-Guided Retriever (GGR) uses a 2,859-node knowledge graph as a candidate gate with intent routing to deliver 71.6x token reduction. The Scene-Aware Ranker (SAR) applies 19-class entity recognition and explicit scenario annotations; negative knowledge alone contributes 25 percentage points of Hit@10 gain. On two 100-question benchmarks, Hit@10 rises from 19.1% to 96.6% (+77.5pp) and knowledge coverage from 56% to 77%, at 4.84--5.33s end-to-end latency.
Deep research over data lakes requires an LLM agent to investigate evidence across thousands of heterogeneous tables and passages to synthesize a report. Existing methods perform iterative retrieval and generation, letting accumulated context determine what to investigate next, which can overexploit locally promising evidence and fail to cover distinct semantic regions under a fixed budget. To address this, we cast deep research over data lakes as a budgeted search problem and present Baikal - a framework that clusters heterogeneous evidence into semantic regions, then searches over them adaptively to balance exploration and exploitation. Within each selected region, Baikal generates and investigates region-grounded subquestions, using finding quality as rewards to update region-level value estimates and guide search under policies ranging from random and LLM-guided selection to Bayesian $ε$-greedy and UCB. We evaluate Baikal on 15 queries each over HybridQA and TAT-QA data lakes containing 10,993 and 2,757 tables, respectively, together with 227K Wikipedia passages and 13K financial report passages. We assess research quality with a new rubric covering groundedness, relevance, diversity, and utility, and use GPT-5-mini to score Baikal and strong baselines, including DeepSearcher and an OpenCode research agent with retrieval and clustering variants. Across both data lakes, Baikal performs strongly under several region-selection policies; its best configuration improves report scores over the strongest baselines by 28% on HybridQA and 36% on TAT-QA. Our analyses attribute these gains to organizing and exploring semantic evidence regions, which improves groundedness and diversity and yields more useful findings under the same subquestion budget. These results demonstrate the value of structured semantic exploration for systematic research and discovery over heterogeneous data lakes.
Analyzing financial documents such as 10-K filings, tabular disclosures, and macroeconomic reports demands expert reasoning and extensive time. However, existing Retrieval-Augmented Generation systems often struggle to process hybrid text-table structures or the massive scale of financial documents. To address these challenges, we propose Hierarchical Reranker, a RAG framework designed to improve retrieval performance and generative reliability across large-scale financial datasets. The system integrates three key innovations: Pre-Retrieval Optimization, enhancing query clarity and search efficiency through normalization, keyword expansion, and table transformation; Hierarchical Reranker Architecture, improving retrieval precision through a two-stage ranking mechanism; and Long-Context Management, preserving reasoning accuracy through adaptive input partitioning and fusion under extensive contexts. Across multiple benchmarks, including FinQA, FinanceBench, and ConvFinQA, the proposed system achieved an NDCG@20 score of 0.7918 and demonstrated superior factual consistency. Its robustness was further validated by achieving second place in the ACM-ICAIF '24 FinanceRAG Challenge. This work presents a deployable, domain-optimized RAG pipeline that enhances both the accuracy and scalability of financial reasoning, paving the way for automated audit reporting and quantitative investment analysis. The source code will be made publicly available on GitHub upon acceptance.
In modern recommendation systems, retrieval serves as a primary stage responsible for filtering billions of candidate items down to thousands prior to refined ranking. To make this massive search effective and efficient, the system relies on ranking accuracy and indexing efficiency. However, these two objectives are traditionally misaligned: while the former optimizes for the alignment between ranking predictions and user behavior, the latter optimizes for a structural grouping of item representations which enables fast search among billions of candidates. Thus, despite extensive efforts to scale up interaction modeling for retrieval, they remain fundamentally limited by the structural misalignment between the ranking objectives and the proximity-learned index. In this work, we address this long-standing dichotomy by proposing a new holistic retrieval framework, OneShot. It is an end-to-end, in-model index learning framework that natively aligns index learning with ranking objectives. Using this joint learning as a structural foundation, OneShot pushes the boundaries of retrieval expressiveness by scaling interaction modeling with neural scoring beyond the persistent dot-product bottleneck. OneShot is fully deployed in Instagram's industrial short-video recommendation system, driving significant wins in user daily sessions, engagement, and time-spent. Additionally, OneShot achieves a $20\%$ recall gain at the operational ranking volume and a 10x efficiency improvement at an equivalent recall level.
Traditional search systems are optimized to retrieve items that strictly match a query, often prioritizing precision over recall. In e-commerce marketplaces and particularly grocery, this paradigm is limiting, as user satisfaction and commercial outcomes depend heavily on the discoverability of substitute, complementary, and thematically related items. In this paper, we present a scalable system for discovery-augmented search that leverages intent-conditioned recall expansion. Our approach generates implicit user intents to expand candidate recall while maintaining relevance. The system addresses the cost-quality tradeoff of generative retrieval through a two-stage hybrid architecture. First, we leverage closed-weight large language models (LLMs) to maximize discoverability for head queries. To extend these benefits to tail queries, we then introduce a finetuned small language model (SLM), trained via LoRA adapters and teacher-student distillation. We evaluate the system using a rigorous dual framework: (a) LLM-as-a-judge metrics validated against human preferences for semantic quality, and (b) end-to-end session-level purchase analysis. Results demonstrate that our approach improves both intent generation quality and downstream retrieval effectiveness, extending discovery coverage from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost, offering a viable path for deployment in large-scale marketplaces. Beyond relevance gains, discovery-augmented search may serve as a marketplace-balancing mechanism, giving long-tail and emerging supply an opportunity for query-conditioned exposure.
Sungguk Cha, DongWook Kim, Mintae Kim +3cs.CV cs.CL cs.IR
Finding a long document relevant to a multi-part request is not the same as establishing that it contains every requested piece of evidence. We study this gap for conjunctive document retrieval, where two or three explicit conditions must be supported on different pages of one document. We use n-Clue as a controlled measurement instrument: 1{,}000 queries over 2{,}021 documents pair all-condition golds with naturally occurring documents that satisfy only a subset, and a complete-first success requires a top-10 gold to precede every released subset qrel. Across 70 configurations, condition-wise decomposition improves two dense backbones by 6.8--7.3 points and lexical--visual fusion adds 8.7, while four generic rerankers all reduce Gold-NDCG; these directions replicate on a four-source stress set. Scaling one dense family from 0.6B to 8B changes complete-first success by 0.0 points. The strongest displayed hybrid illustrates the resulting gap: it finds a gold for 81.1\% of queries but succeeds complete-first on only 35.8\%, and the gap persists across condition count, target length, candidate density, query rendering, and the four-source stress set. Finally, page-aware visual systems surface stored support for every condition on only 5.1--5.3\% of queries. These results identify condition coverage, rather than gold discovery alone, as the central bottleneck.
Choosing the right text embedding model is one of the most consequential -- and most frequently under-examined -- decisions in building a retrieval or search system, yet the model that tops a leaderboard is rarely the best choice for a given deployment. This report develops a practical, evidence-based framework for embedding model selection, built on a benchmarking study that evaluates T3EM (Text 3 Embedding Model), a commercial API-based embedding model, against a broad set of open-source alternatives on English-language retrieval tasks, and situates these findings within the wider Massive Text Embedding Benchmark (MTEB) landscape spanning classification, clustering, semantic similarity, reranking, pair classification, bitext mining, and summarization. Beyond raw benchmark scores, the report traces the full path from embedding model to retrieved result -- how embeddings are produced, how they are indexed and searched at scale, and how document chunking strategy shapes retrieval quality -- so that model choice can be reasoned about as one decision within a complete retrieval pipeline rather than in isolation. The result is a consolidated set of practical recommendations for selecting an embedding model according to task, latency, cost, and deployment constraints.
Congfei Zhang, Jingxiao Ma, Xiaodong Liu +14cs.IR cs.CL cs.LG
Dynamic Product Ads (DPA) require retrieving relevant items from multi-million product catalogs, balancing two competing objectives: retargeting (re-surfacing known interests) and prospecting (discovering new categories). While Large Language Models (LLMs) capture semantic intent better than traditional embedding models, deploying them at scale introduces prohibitive inference costs and lexical mismatch issues. Through controlled experiments on millions of users, we demonstrate a critical retrieval decomposition: rule-generated queries excel at retargeting on a lexical BM25 index, while LLM-generated queries excel at prospecting on a dense ANN index. Building on this, we propose SMART (SeMantic-aware Adaptive ReTrieval). To manage costs, a lightweight quality gate identifies coverage gaps in initial keyword results, adaptively routing only the ~10% of users who benefit from semantic prospecting to the LLM path. Offline evaluation demonstrates that this gated approach captures the bulk of semantic prospecting gains in Relevance Score while maintaining competitive re-targeting performance at a 90% reduction in LLM costs. Finally, in a 2-week online A/B test at Snap, SMART improved the ad conversion rate by +27.6% over a strong embedding-based baseline.
Traditional conversational recommenders entangle retrieval and response generation within a single text interface, so exact entity cues fade as the dialogue's intent evolves, which compromises explanation credibility. We address this within the ACM RecSys Challenge 2026, which mandates both top-20 ranking and evidence-grounded response generation. This paper presents the third-place solution by team "swyoo" for the Blind-B industry track. We decouple retrieval and response into separate pipelines connected strictly via ranked tracks and metadata. Retrieval combines a hybrid lexical-dense pool for exact matching with a task-adapted pool driven by fine-tuned Qwen 8B adapters. Candidates are calibrated via LightGBM, then routed to an evidence-grounded propose-assign-select (PAS) framework to structure responses. This system also ranked second on the explanation-quality leaderboard in the final blind evaluation. Our findings demonstrate that: (i) isolating retrieval and response preserves both catalog cues and fluid intent; (ii) structuring generation via explicit evidence assignment is key to this near-best-in-class explanation reliability.
Sam O'Nuallain, Nithya Rajkumar, Ramya Narayanasamy +3cs.IR cs.AI cs.CL
We present AutoIndex, a framework for learning representation programs: executable transformations that map raw documents into the representations exposed to a retrieval system. Rather than tuning retrievers, rerankers, or a small set of preprocessing hyperparameters, AutoIndex searches over programs that slice, enrich, normalize, reweight, or reorganize documents before indexing. At each iteration, AutoIndex performs validation-guided program search, in which agents diagnose failures of the current program and synthesize candidate updates, retaining only updates that improve retrieval quality under the resulting index. We evaluate AutoIndex on CRUMB, a benchmark of heterogeneous retrieval tasks, with BM25 held fixed across all experiments. The learned programs improve recall over a static full-document BM25 baseline on all 8 tasks, with average gains of +8.4% in Recall@100 and +8.3% in nDCG@10, and largest gains of +30.5% in Recall@100 and +43.6% in nDCG@10. These results suggest that document representation should not be treated as a fixed preprocessing choice made before retrieval begins, but as an explicit optimization target. Code to reproduce our results is available at https://github.com/auto-index/autoindex.
Nikit Srivastava, Daniel Vollmers, René Speck +3cs.CL
Large language models (LLMs) have demonstrated strong reasoning performance, but their tendency to hallucinate limits their reliability in knowledge-intensive tasks requiring up-to-date and grounded information. Combining knowledge graphs (KGs) with LLMs facilitates the use of explicit symbolic knowledge that can be continuously updated without costly fine-tuning, while benefiting from rapidly advancing LLM reasoning. We propose MARS, a scalable knowledge graph question answering (KGQA) approach that requires no model fine-tuning. Rather than relying on open-ended agentic exploration, MARS performs a structured retrieval procedure that links question entities to the KG and iteratively retrieves relevant next-hop information. At each step, MARS decides whether to continue graph traversal or to generate the final SPARQL query, allowing the model to adapt the retrieval depth to the question while keeping the overall pipeline more predictable than fully agentic approaches. We evaluate MARS on three established KGQA benchmarks across several LLMs and settings, including multilingual evaluation, and provide insights through ablation studies and error analysis. Our approach achieves competitive performance relative to state-of-the-art methods while remaining efficient and scalable. The evaluation results, code and resources are publicly available: https://github.com/dice-group/mars-kgqa.
Retrieval over corpora that mix several domains often returns relevant but wrong-domain evidence that ranking metrics miss and that conformal risk control bounds only marginally, under-covering the worst domains. This work introduces C3R, a drop-in control layer that, from an inferred domain posterior and no query-time label, certifies a per-domain contamination budget where feasible and otherwise abstains rather than silently violating; on the hardest domains it guarantees a reduction, not a tight bound. The core is a two-split scheme built on risk-controlling prediction sets, whose finite-sample transfer bound crosses from the inferred to the true domain with fully estimable slack, supports heterogeneous budgets, and inverts for deployment. Population validity rests on this bound and a controlled simulation; across a thousand resampled calibrations the certificate never violates (a stability result) while marginal control violates the most-contaminated domain in every draw, and soft demotion retains more recall than the strongest calibrated cascade at equal certified contamination. The method replicates across open testbeds including an independent one from public federal regulations, and an LLM-judged downstream probe indicates wrong-authority grounding rises with contamination and falls under control. The layer is frozen-stack and reranker-agnostic.
Large language models (LLMs) are increasingly used as backbone architectures for recommender systems because of their strong sequence modeling and representation learning capabilities. However, most LLM-based recommenders operate primarily on discrete textual tokens, whereas practical recommendation pipelines also rely on continuous numerical features and dense embedding features produced by upstream feature engineering or pretrained encoders. This mismatch limits the ability of LLM-based models to exploit fine-grained non-textual signals. We propose a soft-token fusion framework that maps numerical and embedding features into the LLM embedding space, allowing heterogeneous recommendation signals to be consumed through the standard token interface. We instantiate the framework in a shared-parameter LLM-based two-tower retrieval model and introduce an interaction-based fusion module that refines embedding and numerical soft tokens before they are inserted into the final LLM input. Experiments on three Amazon recommendation benchmarks show that soft-token fusion improves retrieval performance over LLM-based baselines, and that interaction-based fusion is more effective than direct concatenation of heterogeneous soft tokens.
Large language model (LLM)-based agentic search systems are often evaluated as if the underlying LLM were the only component that matters, yet their measured performance also depends on the surrounding search environment: the Wikipedia snapshot, preprocessing pipeline, chunking policy, retrieval backend, tool schema, observation format, and answer submission rule. These details are frequently under-specified, making it difficult to compare results or reproduce reported baselines. We present SimpleWikiSearch, whose corpus construction, retrieval stack, tool contract, and evaluation protocol are explicit and runnable. The environment starts from a full English Wikipedia dump, cleans and chunks the corpus, builds keyword and dense retrieval indexes, and exposes a minimal tool interface consisting of \texttt{search}, \texttt{open\_url}, and \texttt{submit\_answer}. We report baseline results on six QA datasets using open-source LLMs and provide a random-300 subset for comparisons with closed-source commercial models. SimpleWikiSearch provides a domain-specific agent harness and a controlled offline environment for reproducible agentic-search evaluation. Its contribution is this specified reference setup, rather than a new agent algorithm. Code and data will be available at: https://github.com/JimXiongGM/simple_wiki_search.
Mahmoud Hany, Mourad ElSheraey, Mahmoud Said +1cs.CL cs.IR
Answering questions over a set of transactional legal documents is most simply done by injecting the whole corpus into the LLM's context window on every query. That baseline maximises retrieval recall, but its token footprint scales with the corpus rather than the question, and long-context degradation scales with it. We report what it took to replace full-corpus injection in a legal-document analysis system, comparing it against two structured retrieval modes over our proprietary structure-aware chunking: embedding retrieval (NAVEMBED) and LLM navigation over a compact structured index (NAVINDEX). On a 20-question benchmark with verified ground-truth answers, a position-bias-controlled, reference-anchored pairwise judge scored semantic retrieval with reranking tied with injection on 16 of 18 document-bound questions (injection preferred on 2) while attending to 17.3x fewer input tokens (a general-text-embedding (GTE) configuration reaches 29.9x at a lower tie rate); both modes were judged tied on the 2 out-of-scope controls. NAVINDEX was judged tied on all 18 at a 1.61x smaller total token footprint, a ~56x smaller answering context, and 25% lower dollar cost. We derive a closed-form caching-crossover rule: cached injection is cheaper in dollars only while the corpus stays below roughly ten times the retrieval payload. Scope and uncertainty are quantified in Section 8.