Jul 2026· International Conference on the Theory of Information Retrieval· 0 citations· 19 references
Computer Science
TL;DR
This work presents a novel framework that operationalizes explicit reasoning as a control mechanism for LLM-based relevance judgements and systematically quantifies its effects across evaluation settings, suggesting that its benefits depend on task context and complexity rather than being universally beneficial.
Abstract
Relevance judgements are central to information retrieval (IR) evaluation, and there is much interest in using large language models (LLMs) for these judgements. This work presents a novel framework that operationalizes explicit reasoning as a control mechanism for LLM-based relevance judgements and systematically quantifies its effects across evaluation settings. We compare judgements with and without reasoning to assess the impact on accuracy and reliability across multiple datasets and models (e.g., gpt-oss, Gemini 2.5). Our results show that using reasoning often leads to lower-quality judgements and a higher rate of formatting failures, while incurring substantially higher token costs. These findings raise important questions about when and how reasoning should be incorporated into evaluation pipelines, particularly for passage-level relevance assessment, suggesting that its benefits depend on task context and complexity rather than being universally beneficial. This work also contributes a novel dataset of LLM reasoning outputs, augmenting two established test collections to support future research.
Large language models have improved rapidly on tasks with verifiable answers, such as mathematics and programming. Much less is known about their ability to reason about what we call conceptual questions: questions for which no ground truth is realistically accessible and no widely accepted resolution methodology exists, but on which progress can still be made by debating arguments. Most philosophical questions are of this kind, as are central components of questions in AI safety, decision theory, and social choice. Our approach is based on the view that while bottom-line conclusions on such questions are hard to evaluate, individual contextualized arguments can be evaluated far more reliably. We therefore introduce a dataset of 951 argumentative critiques of 442 position texts, spanning topics from AI safety and decision theory to ethics and politics, with 1,458 ratings by six expert raters along dimensions including centrality, strength, correctness, and clarity. We propose two scoring functions and benchmark a range of models. Performance tracks general capability rankings.
Emery Cooper, Caspar Oesterheld, Linh Nguyen et al.· 0 citations
Information retrieval has long focused on ranking documents by semantic relatedness. Yet many real-world information needs demand more: enforcement of logical constraints, multi-step inference, and synthesis of multiple pieces of evidence. Addressing these requirements is, at its core, a problem of reasoning. Across AI communities, researchers are developing diverse solutions for the problem of reasoning, from inference-time strategies and post-training of LLMs, to neuro-symbolic systems, Bayesian and probabilistic frameworks, geometric representations, and energy-based models. These efforts target the same problem: to move beyond pattern-matching systems toward structured, verifiable inference. However, they remain scattered across disciplines, making it difficult for IR researchers to identify the most relevant ideas and opportunities. To help navigate the fragmented landscape of research in reasoning, this tutorial first articulates a working definition of reasoning within the context of information retrieval and derives from it a unified analytical framework. The framework maps existing approaches along axes that reflect the core components of the definition. By providing a comprehensive overview of recent approaches and mapping current methods onto the defined axes, we expose their trade-offs and complementarities, highlight where IR can benefit from cross-disciplinary advances, and illustrate how retrieval process itself can play a central role in broader reasoning systems. The tutorial will equip participants with both a conceptual framework and practical guidance for enhancing reasoning-capable IR systems, while situating IR as a domain that both benefits and contributes to the broader development of reasoning methodologies.
Mohanna Hoveyda, Panagiotis Eustratiadis, A. D. de Vries et al.· Annual International ACM SIG...· 1 citation
Retrieval-augmented generation improves knowledge-intensive question answering, but indiscriminate retrieval can introduce irrelevant evidence and unnecessary computation. We investigate whether verbalized confidence from black-box language models can serve as an actionable signal for retrieval routing. Our method, BeyondUncertainty, first elicits a structured provisional answer and confidence estimate, then applies a model-specific threshold selected on held-out validation data and frozen before test evaluation. Low-confidence questions receive top-5 TF-IDF retrieval followed by a second answer call, whereas high-confidence questions return the provisional answer directly. We evaluate 27,000 policy instances across six QA benchmarks, three model families, and three retrieval policies. BeyondUncertainty achieves 0.483 mean token-level F1, compared with 0.467 for always retrieval and 0.401 for no retrieval, while reducing retrieved passages by 20.4\% relative to always retrieval. When matched on the number of questions routed to retrieval within each dataset-model cell, it outperforms a post-hoc random allocation in 17 of 18 settings, with an average gain of 0.024 F1. Although poorly calibrated as an absolute probability, probe uncertainty modestly predicts question-level retrieval benefit (AUROC = 0.628). However, the additional probe increases total token usage by 28.2\%, revealing a trade-off between more selective evidence acquisition and end-to-end token efficiency.
Chandan Kumar Sah, Xiaoli Lian, Li Zhang· 0 citations
Large language models (LLMs) are increasingly used in information retrieval (IR) pipelines as relevance judges and re-rankers. Yet most analyses remain output-centric, evaluating generated labels or scores while offering limited insight into how relevance is represented inside the model. In this work, we study whether query-document (q-d) relevance is linearly decodable from residual-stream activations in instruction-tuned LLMs, how this signal compares with generated relevance judgments, and whether it transfers across languages. Using the TREC DL20 and MIRACL evaluation collections, we guide medium-scale LLMs (4-9B parameters) with UMBRELA-style relevance judgment prompts, extract last-token activations from every transformer layer, and train linear probes to predict relevance labels. We compare probe predictions with generated judgments and use TREC DL20 to test whether probe-derived pseudo-labels preserve system rankings against human judgments. Our results suggest that q-d relevance is encoded as a depth-dependent signal: probe performance is weak in early layers and strongest in middle-to-late layers, indicating that relevance becomes more linearly accessible after contextual integration. Most importantly, in several models, validation-selected probes match or outperform generated judgments and better preserve system rankings, revealing a separation between internal relevance representation and external expression. Multilingual experiments suggest partial cross-language portability, although transfer remains weaker than within-language decoding. Overall, this work provides a representation-level perspective on LLM-based relevance assessment. Layer-wise probing can help diagnose where relevance emerges, when generated judgments fail to reflect internally available evidence, and how relevance representations vary across languages, datasets, and model families.
Pietro Bernardelle, Samaneh Mohtadi, Stefano Civelli et al.· 0 citations
Lite-CoNER is proposed, a lightweight NER framework that achieves an effective balance between recognition accuracy and inference efficiency and provides a transparent view of the decision-making process, proving that lightweight models can effectively inherit complex logic through structured distillation.
Yang Wang, Lushuang Gao· International Conference on...· 0 citations
The results demonstrate that suppressing entity-linked supervision during pretraining induces a shift in epistemic behavior: KLLMs rely less on parametric knowledge and more on external evidence, leading to improved reliability under realistic conditions, suggesting that pretraining-time control over knowledge acquisition can complement retrieval-augmented and tool-based systems by providing a more evidence-sensitive base model.
Roi Cohen, Yvan Carr'e, Nick Lechtenborger et al.· 0 citations
