PyTorchKR Papers

최신 논문

총 222편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2608.25735v3cs.CR

Pointing the Way, Hiding the Destination: Practical Private Dense Retrieval at Scale

Hosted retrieval-augmented generation (RAG) and semantic search allow users to query valuable provider-held corpora, raising two competing demands: to hide each query and chosen result, yet reveal only the documents that the user is authorized to receive. Existing cryptographic approaches either make this costly by processing the entire corpus for every query, or sacrifice quality for efficiency by scanning a few clusters. We repurpose learned deep hashing as a private filter: a randomized binary code points the provider to a short candidate list, while encrypted reranking and oblivious key transfer protect the precise query and final selection. This shortlist short-circuits full-corpus cryptographic search without sacrificing retrieval quality: with 200-500 candidates, it closely matches full-corpus retrieval across five zero-shot corpora spanning 25K to 5.4M documents. On the full 2.68M-passage NQ corpus over a 10-Gbps link, our protocol only adds 0.38 seconds, or 6.5%, to a 128-token Qwen3-32B RAG pipeline. The released code satisfies directional metric differential privacy (DP) and substantially reduces embedding-inversion and property-inference leakage, demonstrating that a carefully learned shortlist can make private dense retrieval both accurate and practical.

arXiv:2607.02338v3cs.DB

Graph-Based Nearest-Neighbor Search with Recall Guarantees Using Graph Spanners

Graph-based indexes such as Hierarchical Navigable Small World (HNSW) graphs make nearest-neighbor retrieval efficient, but a high average recall can hide queries whose results fall below an application's quality requirement. We introduce Certify-then-Rectify (CTR), which returns an initial graph-search result when a calibrated rule accepts it and searches for missing neighbors otherwise. The recovery search reuses the existing graph and the distances already evaluated by the initial search. To determine how far this search should proceed, we relate metric distances to graph-path lengths through graph stretch. We develop Stretch-Bounded Expansion from Query (SBE-Q), together with Metric Bound Verification (MBV) to prune unnecessary work, and estimate its budget from paths to the fraction of true neighbors needed for the requested recall. Finite-sample tolerance bounds support this recall-aligned estimate; paths recorded during proposal recovery offer a conservative alternative to separate shortest-path computations. Independent tests certify initial acceptance and recovery outcomes, giving an end-to-end population recall guarantee even when the estimated stretch does not cover every relevant path. If recovery cannot be certified, rejected queries use exact scan. Experiments on HNSW compare five stretch estimators, assess recovery efficiency, and evaluate compliance and throughput on databases of up to 100 million vectors. CTR is not specific to HNSW, and we also evaluate it on another graph index.

arXiv:2511.16543v3cs.IR

The Oracle and The Prism: A Decoupled and Efficient Framework for Generative Recommendation Explanation

The integration of Large Language Models (LLMs) into explainable recommendation systems often leads to a performance-efficiency trade-off in end-to-end architectures, where joint optimization of ranking and explanation can result in suboptimal compromises. To resolve this, we propose Prism, a novel decoupled framework that rigorously separates the recommendation process into a dedicated ranking stage and an explanation generation stage. This decomposition ensures that each component is optimized for its specific objective, eliminating inherent conflicts in coupled models. Inspired by knowledge distillation, Prism leverages a powerful, instruction-following teacher LLM (FLAN-T5-XXL) as an Oracle to produce high-fidelity explanatory knowledge. A compact, fine-tuned student model (BART-Base), the Prism, then specializes in synthesizing this knowledge into personalized explanations. Our extensive experiments on benchmark datasets reveal a key finding: the distillation process not only transfers knowledge but also acts as a noise filter. Our 140M-parameter Prism model significantly outperforms its 11B-parameter teacher in human evaluations of faithfulness and personalization, demonstrating an emergent ability to correct hallucinations present in the teacher's outputs. While achieving a 24x speedup and a 10x reduction in memory consumption, our analysis validates that decoupling, coupled with targeted distillation, provides an efficient and effective pathway to high-quality, and perhaps more importantly, trustworthy explainable recommendation.

arXiv:2611.17030v1cs.CL

Similar but Not Redundant: Embedding-Gated Refinement for High-Granularity Topic Models

High-granularity topic modelling extracts hundreds of niche topics from a corpus, yet this regime is rarely studied and is more commonly framed as a failure mode: "over-clustering," in which coherent themes fragment into redundant subtopics. We argue that this framing conflates genuine redundancy with topics that are semantically similar but meaningfully distinct, and that the central challenge in high-granularity modelling is separating the two. Recent post-hoc refinement methods address topic redundancy through either embedding-based similarity thresholding or LLM-guided merging. We show that neither suffices in high-granularity settings and propose embedding-gated iterative merge (EGIM), a hybrid approach that iteratively generates context-rich topic representations, uses embedding similarity to propose candidate merges across the topic space, then applies LLM judgement to accept only genuinely redundant pairs while preserving niche distinctions. Across two patent corpora and one legal corpus, LLM judgement rejects 83% of embedding-identified candidates as non-redundant. EGIM improves discriminability on merged topics by 25-55% over a prior iterative LLM baseline while better preserving the original topic distribution.

arXiv:2611.17226v1cs.IR

Sheaf-Inspired Structured Scoring Extensions of LightGCN for Collaborative Filtering

Graph-based collaborative filtering learns user and item representations by propagating embeddings over a user--item interaction graph. We investigate a modular, sheaf-inspired scoring extension placed on top of a three-layer LightGCN backbone. The motivation comes from cellular sheaves, where local vector spaces are related by linear restriction maps; here, this idea motivates structured, pair-dependent transformations of user and item embeddings rather than a change to graph propagation. We compare the identity inner-product scorer, diagonal, rank-one low-rank, and full matrix residual scorers, and a parameter-matched residual MLP baseline (\texttt{mlp\_full}) on Amazon-Book, Gowalla, MovieLens-1M, and Yelp2018. We report ranking quality and catalog coverage at cutoffs 10 and 20. The selected structured variants show modest, dataset-dependent changes in ranking quality and sometimes substantial changes in coverage. Our method is inspired by sheaf geometry but does not construct a cellular sheaf or implement sheaf-Laplacian diffusion; the experiments therefore do not establish a causal benefit from sheaf geometry itself.

arXiv:2611.16060v1cs.DB

REWEAVE: Online Learning for Workload-Adaptive HNSW Indexes

Vector databases rely on Hierarchical Navigable Small World (HNSW) indexes for similarity search, but a graph built from data alone can be poorly matched to a skewed or changing query workload. We present REWEAVE (Reward-Guided Rewiring of Vector Graphs), an online index-maintenance method that learns edge utility from query feedback and periodically rewires the bottom layer within the original per-node degree cap. The method supports sparse exact-neighbor feedback combined with self-labels, and standard HNSW serving without learning overhead once adaptation is disabled. Under explicit assumptions we prove value-estimation and graph-stabilization guarantees. Under stationary skew, REWEAVE improves held-out recall by up to 23.3 percentage points over HNSW. In comparisons with query-aware baselines on up to 100M vectors, it achieves up to 3.3x NGFix's query throughput at recall@100 of 0.80. On Amazon-ESCI with query drift, it increases stream recall@100 from 0.745 to 0.979 with 34% fewer search distance computations than frozen HNSW. These results illustrate the potential to improve retrieval quality and reduce search effort within an existing graph-degree budget, while incurring additional work during adaptation.

arXiv:2611.16148v1cs.DB

Calibrated Early Termination with Bounded Optimality Gaps for Constrained Vector Search

Vector search over learned embeddings is the retrieval layer of recommendation and retrieval-augmented generation, and many applications cap how many answer items one category may contribute. Graph indexes answer a query by best-first traversal, and where it stops determines its cost and whether its answer is exact. We study early termination for such constrained $k$-nearest-neighbor ($k$-NN) queries, with plain $k$-NN as a special case, and seek the cheapest stopping rule whose answer is exact with probability at least $1-\varepsilon$. Learned early-termination methods instead target average recall, which does not guarantee exact answers. Multiple testing can calibrate any parameterized stopping rule to our target, but it cannot tell how much a better rule with the same risk and observations would save. Answering this needs a lower bound on the optimum of a rule family too large to enumerate, valid for the query population. Our key idea is to declare the observations a rule may use and to consider every rule monotone in them. Risk-constrained stopping then becomes a linear program over an order polytope, whose Lagrangian minimizers form a nested chain of staircase rules computed exactly by dynamic programming. Fixed-sequence binomial tests along this chain calibrate our rule, CaST, whose failure probability is at most $\varepsilon$ with confidence $1-\delta$. Lagrangian weak duality bounds the optimum of the declared class from below at all risk levels at once, without a union bound over the rules. On two datasets of one million items each, CaST replayed on logged searches costs at most 1.132 times the class optimum for risk levels from 0.03 to 0.10, at 95% joint confidence per setting. Larger savings thus need a rule outside the class. At a common risk target on held-out queries, CaST is faster than every other calibrated rule we time, which take 4.8% to 45.5% more mean time.

arXiv:2611.16425v1cs.IR

RMIT-ADM+S at TREC RAG 2026: Research Agents, Retriever Ablations, and Open-Weight Replication

We describe the RMIT-ADM+S team's participation in the TREC 2026 Retrieval-Augmented Generation (RAG) track using the ClimbMix corpus. We submitted seven research-agent runs using proprietary OpenAI and open-weight models, exploring different retrieval strategies, writer models, and agent workflows. Pre-submission ablations identified writer selection and adjacent-chunk expansion as the main factors affecting answer quality. In the official NIST evaluation, six of seven runs exceeded the track median across all three rubric/nugget collections. Our best-performing run, brief-hybrid, achieved the highest all-item coverage among our submissions and tied the track maximum (0.892) for Batch-1 vital coverage. Human side-by-side preferences differed across evaluation batches: the open-weight owa-glm5 led Batch 1 despite lower automated coverage, while sol-research-first led Batch 2. These findings highlight differences between automated coverage measures and human preferences, as well as the potential of lower-cost open-weight models for research-agent systems.

arXiv:2611.14938v1econ.GN

When Does a Recommendation Spark Curiosity? Theory, Measurement, and Field Evidence

Curiosity, the drive to close the gap between what one knows and what one wants to know, pulls consumers beyond their familiar content, sustains engagement, and breaks them out of the filter bubbles of repetitive, relevance-only recommendations. Despite its importance, curiosity is hardly incorporated into modern recommender systems for lack of a formal measure. The information-gap theory of curiosity provides no formal quantitative model that a recommender system can operate on, and existing proxy metrics do not accurately measure the curiosity that a recommended item evokes. To fill this research gap, we formalize the information-gap theory into five axioms of curiosity and construct the canonical formulation of a computational measure of the curiosity that a recommended item evokes in a consumer. We validate our curiosity measure through a survey-based study, in which it matches consumers' self-reported curiosity and outperforms alternative proxy metrics. We subsequently build a curiosity-aware recommender system that incorporates this measure into the utility function design, and we show through offline experiments on three real-world datasets that it significantly outperforms the state-of-the-art methods. Furthermore, we conduct a large-scale online controlled experiment at a major video-streaming platform in Asia, where we show that our method significantly increases click-through rate, video views, and time spent over the latest production system, leading to tangible economic impact for the company. Our method has since been deployed on the platform and remains in operation. Our findings establish consumer curiosity as a measurable and actionable objective for recommender systems, with important managerial implications for digital platforms to recommend what consumers wonder about.

arXiv:2611.13972v1cs.IR

Data Quality Challenges in Multimodal Tourism Recommender Systems

Modern recommender systems increasingly consider diverse data modalities to enhance model performance. However, the quality of the underlying multimodal data (e.g. data from social media platforms) has received limited attention so far. Focusing on tourism recommendation, we examine real-world challenges encountered when using datasets from Yelp and Instagram. Issues include noise from fake reviews, hallucinated content from LLMs, modality contradictions, and demographic bias. We outline future directions such as quality-aware fusion, structured and controlled LLM-based content generation to reduce hallucinations, and user group differentiation to improve the robustness and reliability of multimodal recommendation systems.

arXiv:2611.13966v1cs.IR

Controlled Knowledge Engineering for Taxonomy Induction with Tool-Grounded LLM Agents and Runtime-Adaptive Policies

Taxonomies are structured knowledge representations that support semantic search, recommendation, product catalog organization, and retrieval-augmented generation. Their automated construction from a fixed set of candidate entities requires more than plausible text generation: the resulting hierarchy must be complete, structurally valid, and semantically coherent. Large Language Models (LLMs) are promising for this task, but iterative free-form generation remains vulnerable to omitted or hallucinated entities, invalid hierarchical relations, and inconsistent organizational principles. We introduce TIGA (Taxonomy Induction with Tool-Grounded Agents), a knowledge-engineering architecture that formulates taxonomy induction as an iterative decision process over an explicit state and constrained action space. TIGA combines executable taxonomy-editing tools, an memory that records tool calls, rationales, and outcomes, and a runtime-adaptive organizational policy that the agent revises during inference. An optional global reviewer evaluates the constructed taxonomy against the candidate set and the active policy. We evaluate TIGA on four benchmark taxonomies using two LLM backbones and compare it with single-shot and iterative prompting-based baselines. TIGA achieves the strongest structural-quality results in most evaluated configurations, but does not dominate on every precision measure. The gains are accompanied by substantially higher token consumption, API-call volume, and inference time. Ablation results show consistent benefits from the policy and interaction-memory components, whereas the effect of global review depends on the dataset and the available revision budget. These findings characterize tool grounding and runtime policy adaptation as promising but costly mechanisms for improving the controlled construction of structured knowledge representations.

arXiv:2611.13914v1cs.AI

From Counterfactual Edits to Human Narratives: LLM-Driven Interpretations for Recommender System Explanations

Recommender systems increasingly influence what users discover and consume online, yet the decision logic behind their outputs often remains opaque. Counterfactual explanation (CFX) methods provide an effective way to show which changes to a user's interaction history would alter a recommendation. In recommendation settings, however, CFXs often require many edits, producing long, low-level lists that are hard for users to interpret. Moreover, most evaluations emphasize edit counts while largely overlooking whether the resulting CFXs are coherent and actionable for users. We argue that this gap between generating valid CFXs and communicating model behavior stems from the lack of user-centered abstraction and limited qualitative evaluation. To bridge this gap, we investigate large language models as mediators that summarize raw CFX interaction edit lists into concise natural language explanations describing higher-level preference patterns. We optimize these mediators for correctness, informativeness, and linguistic quality through preference optimization, and evaluate the resulting explanations with human-calibrated metrics derived from separate human-labeled datasets. On the MovieLens-1M dataset, across multiple counterfactual generators and language models, the resulting summaries achieve high overall linguistic quality, while the strongest semantic results are obtained with high-success CFX methods such as ACCENT and SPINRec. Direct Preference Optimization improves the semantic metrics in almost all evaluated settings.

arXiv:2611.14004v1cs.IR

LUMIR: Decoupled Multimodal Representation Learning from Heterogeneous Data for Next-POI Recommendation

As multimodal tourism-related information collected from heterogeneous data sources has become increasingly available, learning effective representations has attracted attention for improving recommendation performance. Existing approaches often tightly couple multimodal fusion with the architecture and optimization objective of a specific downstream model, which limits the generality and reusability of the learned representations. To address these limitations, we propose LUMIR, a decoupled multimodal representation learning framework based on contrastive learning. Multimodal information collected from heterogeneous data sources is first semantically unified into a shared textual representation space using large language models (LLMs). An interaction-aware fusion mechanism is then designed to model cross-modal interactions, while sequential signals are incorporated as a structural constraint to guide representation learning. To evaluate the learned representations, experiments are conducted on a multimodal point-of-interest (POI) dataset under the next-POI recommendation setting across multiple sequential recommendation backbones. The results show that LUMIR consistently outperforms existing multimodal fusion baselines in both accuracy-oriented and beyond-accuracy metrics, demonstrating that decoupling representation learning from downstream optimization provides an effective paradigm for learning reusable multimodal representations. The code is publicly available at https://github.com/wzehui/LUMIR.

arXiv:2611.14325v1cs.CL

Point-in-Time Encoders for Earnings Calls

Text encoders turn earnings calls into vectors for search, for language models that answer from retrieved passages, and for forecasting models. A live retriever can use an encoder trained once on all available text, as finance retrievers are, but a backtest needs an encoder that read no text written after the date of each vector. To address this look-ahead bias, from 412,248 calls we train vintages under four pretraining recipes, with yearly cutoffs from 2010 to 2024, and each vintage reads only text dated on or before its cutoff. We then fine-tune each vintage on the question-answer pairs of the calls themselves. The vintages we train on calls alone, with 33.8M parameters, retrieve answers on held-out calls better than the general retrievers we tested in every test quarter. Their vectors also raise by about half the rank correlation between next-day returns and forecasts built from past returns and firm size. When each vintage continues the previous one, a later vintage retrieves better on earlier quarters than the vintage available at the time, and a point-in-time language model that answers from the passages the later vintage retrieves inherits the gain. The 2024 vintage of our families also forecasts earnings surprises better, although the interval of this gain includes zero on calls that no vintage read. It forecasts returns no better, so point-in-time vintages matter most for retrieval and little for forecasts of returns.

arXiv:2611.14236v1cs.CL

StaRSE: Compact Russian Sentence Embeddings with a Sign-Coded Static Encoder

Transformer sentence encoders provide strong retrieval quality, but their inference cost becomes a bottleneck when embeddings must be produced on CPUs or refreshed frequently. We introduce StaRSE, a compact Russian static sentence encoder that removes Transformer inference at deployment time: token directions are stored as sign bits, token norms are stored separately, and sentences are encoded by mean pooling over norm-scaled sign directions. StaRSE is initialized from ruBERT-base, fine-tuned on Russian adjacent-sentence and task-shaped pairs, and post-trained with a contrastive stage that directly optimizes the sign-projected token table. On RuMTEB, StaRSE reaches a task mean of $51.16$, $7.98$ points above the strongest included static baseline. The packaged model occupies $11.3$ MiB and has a mean CPU latency of $0.08$ ms per sentence, compared with an estimated $55.7$ MiB 16-bit parameter footprint and $1.43$ ms mean CPU latency for the closest-quality contextual encoder. Ablations examine adjacent-sentence supervision, token norms, and direct sign-projected post-training.

arXiv:2611.14477v1cs.IR

FERN: Cross-Benchmark Table Retrieval for Text-to-SQL with Frozen Retrievers and Learned Ranking

Text-to-SQL systems over large pooled catalogs must first retrieve the tables a question needs. We compare against the published table recall of RASL, which uses Claude 3.5 models to extract keywords and select tables, on pooled BIRD, Spider, and Fiben catalogs. FERN (Feature-Engineered Ranking with No generative LLM) instead fuses three frozen retrievers into 100 candidates, describes each with up to 120 features, and reorders them with LightGBM rankers fitted once on BIRD's training questions. On Spider, where it was not fitted, its candidate pool alone reaches RASL's published recall at fifteen tables (98.0), and the ranker raises the share of questions with every gold table in the top five (complete@5) from 83.9% to 95.6%. At five tables, FERN's recall is within one point of RASL's on Spider (96.7 vs 97.0) and on BIRD with its evidence hints (96.6 vs 97.5), and 8.2 points below on BIRD without them (89.3). On 121 real BEAVER questions, without BEAVER fitting, its recall at fifteen (82.1) falls between two BEAVER-trained retrievers (80.8, 84.5), but its complete@15 is below both (49.6 vs 54.5 and 62.0). On Fiben, it falls far short at fifteen tables (54.2 vs 69.2, different question sets), although its recall at fifty is numerically close to RASL's question-only retrieval result (80.7 vs 79.4). On BIRD dev with evidence, non-retrieval features raise complete@5 by 7.7 points over retrieval signals alone. On BIRD dev, median latency is 208 ms per question on one A100, without hosted model calls.

arXiv:2611.13595v1cs.CL

IFRS Compliance Verification: An Open Multi-LLM Benchmark and Content-Conditioned Ensemble Classifier

Verifying corporate financial disclosures against the International Financial Reporting Standards (IFRS) is a core auditing task in which experts cross-reference disclosures with hundreds of fine-grained regulatory paragraphs: a laborious, costly, and error-prone process that motivates growing interest in large language models (LLMs) as compliance assistants. Yet single-model accuracy remains insufficient for high-stakes use, and naive ensembling is vulnerable to systematic per-model biases that can drag aggregate accuracy below that of the strongest individual annotator. This gap is addressed with a multi-stage framework, supported by a new benchmark of 19,998 IFRS regulation-disclosure pairs comprising 11,161 unanimous multi-LLM labels and 8,837 human annotations on items where the LLMs disagreed. The framework couples a three-stage domain-adapted sentence embedding pipeline (BERT masked language modeling on SEC EDGAR filings, unsupervised TSDAE and SimCSE adaptation, and supervised contrastive alignment) with a learned ensemble classifier that fuses frozen embeddings and LLM probability features from GPT-4.1, Claude Sonnet 4, and Gemini 2.5 Pro via Tree-structured Parzen Estimator (TPE) optimized gradient-boosted trees. On the disagreement set, the ensemble achieves 81.7% weighted F1, exceeding the strongest single LLM by 10.4 points and majority voting by 14.6 points, and outperforms eleven unsupervised aggregation baselines; the retrieval pipeline attains Recall@15 of 45.3% on unseen regulations. The dataset, model checkpoints, annotation platform, and classifier implementation will be publicly released.

arXiv:2611.13539v1cs.IR

Exploiting Feature-invariant Scaling Laws for Efficient A/B Experimentation in Recommendation Systems

Researchers and engineers improve recommendation systems by performing A/B experiments with new model architectures, optimization methods, and data features. However, running full-scale experiments (e.g., full-model training with online data) for every feature change is prohibitively computationally intensive, while the result from a single small-scale experiment is often misleading. To address this challenge, we show that modern recommendation models exhibit feature-invariant scaling: the change of a model's performance follows a predictable scaling-law even when different feature changes are applied to the baseline model. We introduce an efficient method that exploits scaling-laws and uses statistical inference to predict (with confidence intervals) the full-scale effect of an A/B experiment in recommendation systems. This enables practitioners to efficiently filter out neutral or regressive feature changes with only small-scale experimentation. We validate our method on three distinct recommendation model families---Wukong, HSTU, and LLM-based recommendation models---across more than 40 architectural, optimization, and data changes. Our method accurately distinguishes beneficial feature changes from neutral and regressive ones while using only a fraction of the compute required for full-scale evaluation. It also identifies several changes that improve HSTU's state-of-the-art performance on MovieLens-20M. Our results demonstrate that scaling laws can be practically exploited to make compute-efficient decisions for improving industrial-scale recommendation systems.

arXiv:2611.13361v1cs.AI

A Technique to Model Product Information and User Reviews (MPIUR)

Modern e-commerce platforms naturally provide three complementary modalities: textual item descriptions, item images, and user reviews. However, existing multimodal recommendation methods often omit reviews or do not integrate all three modalities within a unified framework. In this paper, we propose MPIUR, a multimodal graph-based recommendation model that encodes reviews in heterogeneous graphs, aligns image and text representations, and adapts collaborative--semantic alignment to item interaction density. Experiments on the Amazon Reviews 2023 Musical Instruments and Baby categories show that MPIUR achieves the best global nDCG@10 and Recall@10 among the evaluated methods on both datasets. It also obtains the best nDCG@10 and Recall@10 on sparse items, indicating that jointly exploiting reviews and multimodal item content is particularly useful when interaction histories are limited.

arXiv:2611.13127v1cs.AI

Activation-Guided Graph Retrieval for Cognitive Memory Reconstruction in Language Model Systems

Retrieval-augmented language model systems generally select context by query-conditioned relevance. We argue that long-lived AI systems require a second operator: activation-conditioned memory availability. We specify an architecture in which request and goal cues seed an authorized memory graph, activation and fan attenuation allocate a bounded traversal frontier, retrieved entities recursively emit cues, and a typed evidence graph is reconstructed before generation. The specification yields five architectural consequences about candidate selection, traversal control, graph regularization, evidence reconstruction, and reinforcement. We derive implementation-level cost estimates and define ablations, failure conditions, and a reproducible evaluation protocol. The central claim is narrowed or falsified if a simpler graph reranker matches full traversal on causal and temporal recovery under equal budgets. This architecture-and-analysis preprint reports no benchmark measurements. If the claim holds, retrieval returns the evidence path that makes a memory available, not only a ranked list.