arXiv subjects

Harrie Oosterhuis

Publications and source records attributed to Harrie Oosterhuis.

At least 19 recordsLinked to original sources

arXiv2026

Revisiting N2DCG: An Empirically Grounded Reformulation of Carousel Recommendation Evaluation

Carousel interfaces have been widely used in video and music streaming services, yet it remains unclear how to properly evaluate recommender systems in these two-dimensional layouts. N2DCG has been proposed to address this gap by adapting NDCG to carousel-based recommendation, but it relies on unverified assumptions borrowed from the single-list web-search setting that do not transfer well to two-dimensional carousel layouts. We identify two substantial limitations of N2DCG: its ideal ranking, used for normalization, violates carousel constraints, and its discount function does not reflect user browsing behavior observed in empirical data. To address both limitations, we propose a reformulation of N2DCG that normalizes appropriately by respecting constraints and uses an empirically grounded discount function. We validate the proposed metric, showing that it better reflects users' empirical behavior on real-world eye-tracking data and better predicts the comparison results of carousel layouts simulated based on empirical examination patterns.

cs.IR↗

arXiv2026

An Epistemic Position-Based Click Model: From Interactions to Epistemic Distributions of Relevance and Bias

User interactions with rankings are affected by both items' relevances and display positions. Accordingly, click probabilities are often modeled as a product of relevance and position factors; and for improving recommendation and search, one needs to disentangle relevance from position bias. However, existing click models only provide frequentist point-estimates that do not capture any measure of epistemic uncertainty. Consequently, there is no indication of how much confidence one should have in their predictions. In this work, we introduce the first evidential deep-learning approach to form an epistemic alternative to the important position-based click model. Our learned model takes as input item and position features and outputs a beta-distribution for every relevance and position-bias variable of the position-based model. These distributions capture epistemic uncertainty about click probabilities and the underlying effects of attraction and position bias. The main challenge of our approach is its optimization for which we propose approximation and conditioning techniques to provide numerical stability and variance reduction. Our experiments indicate that our approach captures epistemic uncertainty in predictions on previously-unseen data, whereas standard policy gradients fail to learn meaningful distributions. We believe our contribution of the first contextual epistemic click model constitutes an important step in incorporating Bayesian uncertainty into click modeling.

cs.IR↗

arXiv2026

Exposure-Based Reinforcement Learning to Rank

Reinforcement learning (RL) methods for learning-to-rank (LTR) can optimize (almost) any ranking goal, e.g., from precision or discounted cumulative gain to fairness-of-exposure or ranking distillation. However, standard RL is ineffective and computationally costly due to the enormous action space in LTR settings. Existing methods reach computational efficiency through custom gradient computation algorithms, but they are very complex to implement and often clash with auto-differentiation. Consequently, existing RL for LTR is not attractive to many practitioners. We reconsider RL for LTR while actively avoiding reliance on custom gradients. Contrary to the existing approaches, we focus on variance reduction and GPU computation. In doing so, we discover that high sample-efficiency can be reached through baseline corrections and partial marginalization. Furthermore, we propose an abstraction that places gradient estimation behind a document-exposure distribution, this enables seamless plug-and-play integration with auto-differentiation. Thereby, one only has to implement a loss as a differentiable function of exposure and RL for LTR can optimize it using auto-differentiation. Our experimental results reveal that our new exposure-based RL for LTR approach converges considerably faster and at significantly higher ranking performance than existing custom gradients, with no additional costs in computation time when using GPUs. In contrast, existing custom gradients result in severe stability issues when converging over many epochs, which never occur for our methods. Thus, we considerably improve RL for LTR methodology by increasing its effectiveness, efficiency, and ease of application.

cs.LG↗

arXiv2026

Following the Eye-Tracking Evidence: Established Web-Search Assumptions Fail in Carousel Interfaces

Carousel interfaces have been the de-facto standard for streaming media services for over a decade. Yet, there has been very little research into user behavior with such interfaces, which thus remains poorly understood. Due to this lack of empirical research, previous work has assumed that behaviors established in single-list web-search interfaces, such as the F-pattern and the examination hypothesis, also apply to carousel interfaces, for instance when designing click models or evaluation metrics. We analyze a recently-released interaction and examination dataset resulting from an eye-tracking study performed on carousel interfaces to verify whether these assumptions actually hold. We find that (i)~the F-pattern holds only for vertical examination and not for horizontal swiping; additionally, we discover that, when conditioned on a click, user examination follows an L-pattern unique to carousel interfaces; (ii)~click-through-rates conditioned on examination indicate that the well-known examination hypothesis does not hold in carousel interfaces; and (iii)~contrary to the assumptions of previous work, users generally ignore carousel headings and focus directly on the content items. Our findings show that many user behavior assumptions, especially concerning examination patterns, do not transfer from web search interfaces to carousel recommendation settings. Our work shows that the field lacks a reliable foundation on which to build models of user behavior with these interfaces. Consequently, a re-evaluation of existing metrics and click models for carousel interfaces may be warranted.

cs.IR↗

arXiv2026

A Simple and Effective Reinforcement Learning Method for Text-to-Image Diffusion Fine-tuning

Reinforcement learning (RL)-based fine-tuning has emerged as a powerful approach for aligning diffusion models with black-box objectives. Proximal policy optimization (PPO) is a popular choice of method for policy optimization. While effective in terms of performance and sample complexity, PPO is highly sensitive to hyper-parameters and involves substantial computational overhead. REINFORCE, on the other hand, mitigates some implementation complexities such as high memory overhead and sensitive hyper-parameter tuning, but has suboptimal performance due to high variance and crucially sample inefficiency, which is the primary notion of efficiency we study in this work. While the variance of the REINFORCE can be reduced by sampling multiple actions per input prompt and using a baseline correction term, it still suffers from sample inefficiency. To address these challenges, we systematically analyze the sample efficiency-effectiveness trade-off between REINFORCE and PPO, and propose leave-one-out PPO ( LOOP), a novel RL for diffusion fine-tuning method. LOOP combines variance reduction techniques from REINFORCE, such as sampling multiple actions per input prompt and a baseline correction term, with the robustness and sample efficiency of PPO via clipping and importance sampling. Our results demonstrate that LOOP effectively improves diffusion models on various black-box objectives, and achieves a better balance between sample efficiency and final performance.

cs.LG↗

arXiv2025

A Non-Parametric Choice Model That Learns How Users Choose Between Recommended Options

Choice models predict which items users choose from presented options. In recommendation settings, they can infer user preferences while countering exposure bias. In contrast with traditional univariate recommendation models, choice models consider which competitors appeared with the chosen item. This ability allows them to distinguish whether a user chose an item due to preference, i.e., they liked it; or competition, i.e., it was the best available option. Each choice model assumes specific user behavior, e.g., the multinomial logit model. However, it is currently unclear how accurately these assumptions capture actual user behavior, how wrong assumptions impact inference, and whether better models exist. In this work, we propose the learned choice model for recommendation (LCM4Rec), a non-parametric method for estimating the choice model. By applying kernel density estimation, LCM4Rec infers the most likely error distribution that describes the effect of inter-item cannibalization and thereby characterizes the users' choice model. Thus, it simultaneously infers what users prefer and how they make choices. Our experimental results indicate that our method (i) can accurately recover the choice model underlying a dataset; (ii) provides robust user preference inference, in contrast with existing choice models that are only effective when their assumptions match user behavior; and (iii) is more resistant against exposure bias than existing choice models. Thereby, we show that learning choice models, instead of assuming them, can produce more robust predictions. We believe this work provides an important step towards better understanding users' choice behavior.

cs.IR↗

arXiv2025

Harnessing Pairwise Ranking Prompting Through Sample-Efficient Ranking Distillation

While Pairwise Ranking Prompting (PRP) with Large Language Models (LLMs) is one of the most effective zero-shot document ranking methods, it has a quadratic computational complexity with respect to the number of documents to be ranked, as it requires an enumeration over all possible document pairs. Consequently, the outstanding ranking performance of PRP has remained unreachable for most real-world ranking applications. In this work, we propose to harness the effectiveness of PRP through pairwise distillation. Specifically, we distill a pointwise student ranker from pairwise teacher labels generated by PRP, resulting in an efficient student model that retains the performance of PRP with substantially lower computational costs. Furthermore, we find that the distillation process can be made sample-efficient: with only 2% of pairs, we are able to obtain the same performance as using all pairs for teacher labels. Thus, our novel approach provides a solution to harness the ranking performance of PRP without incurring high computational costs during both distillation and serving.

cs.IR↗

arXiv2025

Rethinking Click Models in Light of Carousel Interfaces: Theory-Based Categorization and Design of Click Models

Click models are a well-established for modeling user interactions with web interfaces. Previous work has mainly focused on traditional single-list web search settings; this includes existing surveys that introduced categorizations based on the first generation of probabilistic graphical model (PGM) click models that have become standard. However, these categorizations have become outdated, as their conceptualizations are unable to meaningfully compare PGM with neural network (NN) click models nor generalize to newer interfaces, such as carousel interfaces. We argue that this outdated view fails to adequately explain the fundamentals of click model designs, thus hindering the development of novel click models. This work reconsiders what should be the fundamental concepts in click model design, grounding them - unlike previous approaches - in their mathematical properties. We propose three fundamental key-design choices that explain what statistical patterns a click model can capture, and thus indirectly, what user behaviors they can capture. Based on these choices, we create a novel click model taxonomy that allows a meaningful comparison of all existing click models; this is the first taxonomy of single-list, grid and carousel click models that includes PGMs and NNs. Finally, we show how our conceptualization provides a foundation for future click model design by an example derivation of a novel design for carousel interfaces.

cs.IR↗

arXiv2025

Learning to Rank with Variable Result Presentation Lengths

Learning to Rank (LTR) methods generally assume that each document in a top-K ranking is presented in an equal format. However, previous work has shown that users' perceptions of relevance can be changed by varying presentations, i.e., allocating more vertical space to some documents to provide additional textual or image information. Furthermore, presentation length can also redirect attention, as users are more likely to notice longer presentations when scrolling through results. Deciding on the document presentation lengths in a fixed vertical space ranking is an important problem that has not been addressed by existing LTR methods. We address this gap by introducing the variable presentation length ranking task, where simultaneously the ordering of documents and their presentation length is decided. Despite being a generalization of standard ranking, we show that this setting brings significant new challenges: Firstly, the probability ranking principle no longer applies to this setting, and secondly, the problem cannot be divided into separate ordering and length selection tasks. We therefore propose VLPL - a new family of Plackett-Luce list-wise gradient estimation methods for the joint optimization of document ordering and lengths. Our semi-synthetic experiments show that VLPL can effectively balance the expected exposure and attractiveness of all documents, achieving the best performance across different ranking settings. Furthermore, we observe that even simple length-aware methods can achieve significant performance improvements over fixed-length models. Altogether, our theoretical and empirical results highlight the importance and difficulties of combining document presentation with LTR.

cs.IR↗

arXiv2025

RecGaze: The First Eye Tracking and User Interaction Dataset for Carousel Interfaces

Carousel interfaces are widely used in e-commerce and streaming services, but little research has been devoted to them. Previous studies of interfaces for presenting search and recommendation results have focused on single ranked lists, but it appears their results cannot be extrapolated to carousels due to the added complexity. Eye tracking is a highly informative approach to understanding how users click, yet there are no eye tracking studies concerning carousels. There are very few interaction datasets on recommenders with carousel interfaces and none that contain gaze data. We introduce the RecGaze dataset: the first comprehensive feedback dataset on carousels that includes eye tracking results, clicks, cursor movements, and selection explanations. The dataset comprises of interactions from 3 movie selection tasks with 40 different carousel interfaces per user. In total, 87 users and 3,477 interactions are logged. In addition to the dataset, its description and possible use cases, we provide results of a survey on carousel design and the first analysis of gaze data on carousels, which reveals a golden triangle or F-pattern browsing behavior. Our work seeks to advance the field of carousel interfaces by providing the first dataset with eye tracking results on carousels. In this manner, we provide and encourage an empirical understanding of interactions with carousel interfaces, for building better recommender systems through gaze information, and also encourage the development of gaze-based recommenders.

cs.IR↗

arXiv2025

Adaptive Orchestration of Modular Generative Information Access Systems

Advancements in large language models (LLMs) have driven the emergence of complex new systems to provide access to information, that we will collectively refer to as modular generative information access (GenIA) systems. They integrate a broad and evolving range of specialized components, including LLMs, retrieval models, and a heterogeneous set of sources and tools. While modularity offers flexibility, it also raises critical challenges: How can we systematically characterize the space of possible modules and their interactions? How can we automate and optimize interactions among these heterogeneous components? And, how do we enable this modular system to dynamically adapt to varying user query requirements and evolving module capabilities? In this perspective paper, we argue that the architecture of future modular generative information access systems will not just assemble powerful components, but enable a self-organizing system through real-time adaptive orchestration -- where components' interactions are dynamically configured for each user input, maximizing information relevance while minimizing computational overhead. We give provisional answers to the questions raised above with a roadmap that depicts the key principles and methods for designing such an adaptive modular system. We identify pressing challenges, and propose avenues for addressing them in the years ahead. This perspective urges the IR community to rethink modular system designs for developing adaptive, self-optimizing, and future-ready architectures that evolve alongside their rapidly advancing underlying technologies.

cs.IR↗

arXiv2025

Optimizing Compound Retrieval Systems

Modern retrieval systems do not rely on a single ranking model to construct their rankings. Instead, they generally take a cascading approach where a sequence of ranking models are applied in multiple re-ranking stages. Thereby, they balance the quality of the top-K ranking with computational costs by limiting the number of documents each model re-ranks. However, the cascading approach is not the only way models can interact to form a retrieval system. We propose the concept of compound retrieval systems as a broader class of retrieval systems that apply multiple prediction models. This encapsulates cascading models but also allows other types of interactions than top-K re-ranking. In particular, we enable interactions with large language models (LLMs) which can provide relative relevance comparisons. We focus on the optimization of compound retrieval system design which uniquely involves learning where to apply the component models and how to aggregate their predictions into a final ranking. This work shows how our compound approach can combine the classic BM25 retrieval model with state-of-the-art (pairwise) LLM relevance predictions, while optimizing a given ranking metric and efficiency target. Our experimental results show optimized compound retrieval systems provide better trade-offs between effectiveness and efficiency than cascading approaches, even when applied in a self-supervised manner. With the introduction of compound retrieval systems, we hope to inspire the information retrieval field to more out-of-the-box thinking on how prediction models can interact to form rankings.

cs.IR↗

arXiv2024

A Simpler Alternative to Variational Regularized Counterfactual Risk Minimization

Variance regularized counterfactual risk minimization (VRCRM) has been proposed as an alternative off-policy learning (OPL) method. VRCRM method uses a lower-bound on the $f$-divergence between the logging policy and the target policy as regularization during learning and was shown to improve performance over existing OPL alternatives on multi-label classification tasks. In this work, we revisit the original experimental setting of VRCRM and propose to minimize the $f$-divergence directly, instead of optimizing for the lower bound using a $f$-GAN approach. Surprisingly, we were unable to reproduce the results reported in the original setting. In response, we propose a novel simpler alternative to f-divergence optimization by minimizing a direct approximation of f-divergence directly, instead of a $f$-GAN based lower bound. Experiments showed that minimizing the divergence using $f$-GANs did not work as expected, whereas our proposed novel simpler alternative works better empirically.

cs.LG↗

arXiv2024

AQA: Adaptive Question Answering in a Society of LLMs via Contextual Multi-Armed Bandit

In question answering (QA), different questions can be effectively addressed with different answering strategies. Some require a simple lookup, while others need complex, multi-step reasoning to be answered adequately. This observation motivates the development of a dynamic method that adaptively selects the most suitable QA strategy for each question, enabling more efficient and effective systems capable of addressing a broader range of question types. To this aim, we build on recent advances in the orchestration of multiple large language models (LLMs) and formulate adaptive QA as a dynamic orchestration challenge. We define this as a contextual multi-armed bandit problem, where the context is defined by the characteristics of the incoming question and the action space consists of potential communication graph configurations among the LLM agents. We then train a linear upper confidence bound model to learn an optimal mapping between different question types and their corresponding optimal multi-LLM communication graph representation. Our experiments show that the proposed solution is viable for adaptive orchestration of a QA system with multiple modules, as it combines the superior performance of more complex strategies while avoiding their costs when simpler strategies suffice.

cs.CL↗

arXiv2024

Proximal Ranking Policy Optimization for Practical Safety in Counterfactual Learning to Rank

Counterfactual learning to rank (CLTR) can be risky and, in various circumstances, can produce sub-optimal models that hurt performance when deployed. Safe CLTR was introduced to mitigate these risks when using inverse propensity scoring to correct for position bias. However, the existing safety measure for CLTR is not applicable to state-of-the-art CLTR methods, cannot handle trust bias, and relies on specific assumptions about user behavior. We propose a novel approach, proximal ranking policy optimization (PRPO), that provides safety in deployment without assumptions about user behavior. PRPO removes incentives for learning ranking behavior that is too dissimilar to a safe ranking model. Thereby, PRPO imposes a limit on how much learned models can degrade performance metrics, without relying on any specific user assumptions. Our experiments show that PRPO provides higher performance than the existing safe inverse propensity scoring approach. PRPO always maintains safety, even in maximally adversarial situations. By avoiding assumptions, PRPO is the first method with unconditional safety in deployment that translates to robust safety for real-world applications.

cs.LG↗

arXiv2024

Optimal Baseline Corrections for Off-Policy Contextual Bandits

The off-policy learning paradigm allows for recommender systems and general ranking applications to be framed as decision-making problems, where we aim to learn decision policies that optimize an unbiased offline estimate of an online reward metric. With unbiasedness comes potentially high variance, and prevalent methods exist to reduce estimation variance. These methods typically make use of control variates, either additive (i.e., baseline corrections or doubly robust methods) or multiplicative (i.e., self-normalisation). Our work unifies these approaches by proposing a single framework built on their equivalence in learning scenarios. The foundation of our framework is the derivation of an equivalent baseline correction for all of the existing control variates. Consequently, our framework enables us to characterize the variance-optimal unbiased estimator and provide a closed-form solution for it. This optimal estimator brings significantly improved performance in both evaluation and learning, and minimizes data requirements. Empirical observations corroborate our theoretical findings.

cs.LG↗

arXiv2024

Practical and Robust Safety Guarantees for Advanced Counterfactual Learning to Rank

Counterfactual learning to rank (CLTR) can be risky and, in various circumstances, can produce sub-optimal models that hurt performance when deployed. Safe CLTR was introduced to mitigate these risks when using inverse propensity scoring to correct for position bias. However, the existing safety measure for CLTR is not applicable to state-of-the-art CLTR methods, cannot handle trust bias, and relies on specific assumptions about user behavior. Our contributions are two-fold. First, we generalize the existing safe CLTR approach to make it applicable to state-of-the-art doubly robust CLTR and trust bias. Second, we propose a novel approach, proximal ranking policy optimization (PRPO), that provides safety in deployment without assumptions about user behavior. PRPO removes incentives for learning ranking behavior that is too dissimilar to a safe ranking model. Thereby, PRPO imposes a limit on how much learned models can degrade performance metrics, without relying on any specific user assumptions. Our experiments show that both our novel safe doubly robust method and PRPO provide higher performance than the existing safe inverse propensity scoring approach. However, in unexpected circumstances, the safe doubly robust approach can become unsafe and bring detrimental performance. In contrast, PRPO always maintains safety, even in maximally adversarial situations. By avoiding assumptions, PRPO is the first method with unconditional safety in deployment that translates to robust safety for real-world applications.

cs.LG↗

arXiv2024

Local Feature Selection without Label or Feature Leakage for Interpretable Machine Learning Predictions

Local feature selection in machine learning provides instance-specific explanations by focusing on the most relevant features for each prediction, enhancing the interpretability of complex models. However, such methods tend to produce misleading explanations by encoding additional information in their selections. In this work, we attribute the problem of misleading selections by formalizing the concepts of label and feature leakage. We rigorously derive the necessary and sufficient conditions under which we can guarantee no leakage, and show existing methods do not meet these conditions. Furthermore, we propose the first local feature selection method that is proven to have no leakage called SUWR. Our experimental results indicate that SUWR is less prone to overfitting and combines state-of-the-art predictive performance with high feature-selection sparsity. Our generic and easily extendable formal approach provides a strong theoretical basis for future work on interpretability with reliable explanations.

cs.LG↗