@inproceedings{tokarchuk_representation_2026,

title = {Representation Collapse in Machine Translation Through the Lens of Angular Dispersion},

author = {E. Tokarchuk and M. K. Nachesa and S. Troshin and V. Niculae},

editor = {Vera Demberg and Kentaro Inui and Lluís Marquez},

url = {https://aclanthology.org/2026.findings-eacl.126/},

isbn = {979-8-89176-386-9},

year  = {2026},

date = {2026-07-29},

booktitle = {Findings of the Association for Computational Linguistics: EACL 2026},

pages = {2420–2431},

publisher = {Association for Computational Linguistics},

address = {Rabat, Morocco},

abstract = {Modern neural translation models based on the Transformer architecture are known for their high performance, particularly when trained on high-resource datasets. A standard next-token prediction training strategy, while widely adopted in practice, may lead to overlooked artifacts such as representation collapse. Previous works have shown that this problem is especially pronounced in the representation of the deeper Transformer layers, where it often fails to efficiently utilize the geometric space. Representation collapse is even more evident in end-to-end training of continuous-output neural machine translation, where the trivial solution would be to set all vectors to the same value. In this work, we analyze the dynamics of representation collapse at different levels of discrete and continuous NMT transformers throughout training. We incorporate an existing regularization method based on angular dispersion and demonstrate empirically that it not only mitigates collapse but also improves translation quality. Furthermore, we show that quantized models exhibit similar collapse behavior and that the benefits of regularization are preserved even after quantization.},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{de_leon-martinez_recgaze_2025,

title = {RecGaze: The First Eye Tracking and User Interaction Dataset for Carousel Interfaces},

author = {S. Leon-Martinez and J. Kang and R. Moro and M. Rijke and B. Kveton and H. Oosterhuis and M. Bielikova},

url = {https://doi.org/10.1145/3726302.3730301},

doi = {10.1145/3726302.3730301},

isbn = {9798400715921},

year  = {2025},

date = {2025-07-30},

booktitle = {Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval},

pages = {3702–3711},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

series = {SIGIR '25},

abstract = {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.},

note = {event-place: Padua, Italy},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{kang_rethinking_2025,

title = {Rethinking Click Models in Light of Carousel Interfaces: Theory-Based Categorization and Design of Click Models},

author = {J. Kang and M. Rijke and S. Leon-Martinez and H. Oosterhuis},

url = {https://doi.org/10.1145/3731120.3744585},

doi = {10.1145/3731120.3744585},

isbn = {9798400718618},

year  = {2025},

date = {2025-07-30},

booktitle = {Proceedings of the 2025 International ACM SIGIR Conference on Innovative Concepts and Theories in Information Retrieval (ICTIR)},

pages = {44–55},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

series = {ICTIR '25},

abstract = {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.},

note = {event-place: Padua, Italy},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{gregoriadis_large-scale_2025,

title = {A Large-Scale Web Search Dataset for Federated Online Learning to Rank},

author = {M. Gregoriadis and J. Kang and J. Pouwelse},

url = {https://doi.org/10.1145/3746252.3761651},

doi = {10.1145/3746252.3761651},

isbn = {9798400720406},

year  = {2025},

date = {2025-07-30},

booktitle = {Proceedings of the 34th ACM International Conference on Information and Knowledge Management},

pages = {6387–6391},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

series = {CIKM '25},

note = {event-place: Seoul, Republic of Korea},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{nachesa_knn_2025,

title = {kNN For Whisper And Its Effect On Bias And Speaker Adaptation},

author = {M. K. Nachesa and V. Niculae},

editor = {Luis Chiruzzo and Alan Ritter and Lu Wang},

url = {https://aclanthology.org/2025.findings-naacl.369/},

doi = {10.18653/v1/2025.findings-naacl.369},

isbn = {979-8-89176-195-7},

year  = {2025},

date = {2025-07-29},

booktitle = {Findings of the Association for Computational Linguistics: NAACL 2025},

pages = {6636–6642},

publisher = {Association for Computational Linguistics},

address = {Albuquerque, New Mexico},

abstract = {Speech recognition performance varies by language, domain, and speaker characteristics such as accent, but fine-tuning a model on any of these categories may lead to catastrophic forgetting. Token-level k nearest neighbor search (kNN), first proposed for neural sequence decoders for natural language generation (NLG) and machine translation (MT), is a non-parametric method that instead adapts using inference-time search in an external datastore, without training the underlying model. We show that Whisper, a transformer end-to-end speech model, benefits from kNN. We investigate the differences between the speech and text setups. We discuss implications for speaker adaptation, and analyze improvements by gender, accent, and age.},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{ganesh_does_2024,

title = {Does spatio-temporal information benefit the video summarization task?},

author = {A. Ganesh and M. Popa and D. Odijk and N. Tintarev},

url = {http://arxiv.org/abs/2410.03323},

doi = {10.48550/arXiv.2410.03323},

year  = {2024},

date = {2024-07-24},

urldate = {2024-07-24},

booktitle = {AEQUITAS@ ECAI},

abstract = {An important aspect of summarizing videos is understanding the temporal context behind each part of the video to grasp what is and is not important. Video summarization models have in recent years modeled spatio-temporal relationships to represent this information. These models achieved state-of-the-art correlation scores on important benchmark datasets. However, what has not been reviewed is whether spatio-temporal relationships are even required to achieve state-of-the-art results. Previous work in activity recognition has found biases, by prioritizing static cues such as scenes or objects, over motion information. In this paper we inquire if similar spurious relationships might influence the task of video summarization. To do so, we analyse the role that temporal information plays on existing benchmark datasets. We first estimate a baseline with temporally invariant models to see how well such models rank on benchmark datasets (TVSum and SumMe). We then disrupt the temporal order of the videos to investigate the impact it has on existing state-of-the-art models. One of our findings is that the temporally invariant models achieve competitive correlation scores that are close to the human baselines on the TVSum dataset. We also demonstrate that existing models are not affected by temporal perturbations. Furthermore, with certain disruption strategies that shuffle fixed time segments, we can actually improve their correlation scores. With these results, we find that spatio-temporal relationship play a minor role and we raise the question whether these benchmarks adequately model the task of video summarization. Code available at: https://github.com/AashGan/TemporalPerturbSum},

note = {arXiv:2410.03323 [cs]},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{kang_estimating_2024,

title = {Estimating the Hessian Matrix of Ranking Objectives for Stochastic Learning to Rank with Gradient Boosted Trees},

author = {J. Kang and M. Rijke and H. Oosterhuis},

url = {https://doi.org/10.1145/3626772.3657918},

doi = {10.1145/3626772.3657918},

isbn = {9798400704314},

booktitle = {Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval},

pages = {2390–2394},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

series = {SIGIR '24},

abstract = {Stochastic learning to rank (LTR) is a recent branch in the LTR field that concerns the optimization of probabilistic ranking models. Their probabilistic behavior enables certain ranking qualities that are impossible with deterministic models. For example, they can increase the diversity of displayed documents, increase fairness of exposure over documents, and better balance exploitation and exploration through randomization. A core difficulty in LTR is gradient estimation, for this reason, existing stochastic LTR methods have been limited to differentiable ranking models (e.g., neural networks). This is in stark contrast with the general field of LTR where Gradient Boosted Decision Trees (GBDTs) have long been considered the state-of-the-art. In this work, we address this gap by introducing the first stochastic LTR method for GBDTs. Our main contribution is a novel estimator for the second-order derivatives, i.e., the Hessian matrix, which is a requirement for effective GBDTs. To efficiently compute both the first and second-order derivatives simultaneously, we incorporate our estimator into the existing PL-Rank framework, which was originally designed for first-order derivatives only. Our experimental results indicate that stochastic LTR without the Hessian has extremely poor performance, whilst the performance is competitive with the current state-of-the-art with our estimated Hessian. Thus, through the contribution of our novel Hessian estimation method, we have successfully introduced GBDTs to stochastic LTR.},

note = {event-place: Washington DC, USA},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}