Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2511.09329

Sim4IA-Bench: A User Simulation Benchmark Suite for Next Query and Utterance Prediction

Andreas Konstantin Kruff·Christin Katharina Kreutz·Timo Breuer·Philipp Schaer·Krisztian Balog
View original source ↗Read source document ↗Export metadata ↓

Abstract

Validating user simulation is a difficult task due to the lack of established measures and benchmarks, which makes it challenging to assess whether a simulator accurately reflects real user behavior. As part of the Sim4IA Micro-Shared Task at the Sim4IA Workshop, SIGIR 2025, we present Sim4IA-Bench, a simulation benchmark suit for the prediction of the next queries and utterances, the first of its kind in the IR community. Our dataset as part of the suite comprises 160 real-world search sessions from the CORE search engine. For 70 of these sessions, up to 62 simulator runs are available, divided into Task A and Task B, in which different approaches predicted users next search queries or utterances. Sim4IA-Bench provides a basis for evaluating and comparing user simulation approaches and for developing new measures of simulator validity. Although modest in size, the suite represents the first publicly available benchmark that links real search sessions with simulated next-query predictions. In addition to serving as a testbed for next query prediction, it also enables exploratory studies on query reformulation behavior, intent drift, and interaction-aware retrieval evaluation. We also introduce a new measure for evaluating next-query predictions in this task. By making the suite publicly available, we aim to promote reproducible research and stimulate further work on realistic and explainable user simulation for information access: https://github.com/irgroup/Sim4IA-Bench.

Explore related subjects

cs.IR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Andreas Konstantin Kruff, Christin Katharina Kreutz, Timo Breuer, Philipp Schaer, Krisztian Balog. 2026-01-16. Sim4IA-Bench: A User Simulation Benchmark Suite for Next Query and Utterance Prediction. https://arxiv.org/abs/2511.09329

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Do Large Language Models Favor Recent Content? A Study on Recency Bias in LLM-Based Reranking

Large language models (LLMs) are increasingly deployed in information systems, including being used as second-stage rerankers in information retrieval pipelines, yet their susceptibility to recency bias has received little attention. We investigate whether LLMs implicitly favour newer documents by prepending artificial publication dates to passages in the TREC Deep Learning passage retrieval collections in 2021 (DL21) and 2022 (DL22). Across seven models, GPT-3.5-turbo, GPT-4o, GPT-4, LLaMA-3 8B/70B, and Qwen-2.5 7B/72B, "fresh" passages are consistently promoted, shifting the Top-10's mean publication year forward by up to 4.78 years and moving individual items by as many as 95 ranks in our listwise reranking experiments. Although larger models attenuate the effect, none eliminate it. We also observe that the preference of LLMs between two passages with an identical relevance level can be reversed by up to 25% on average after date injection in our pairwise preference experiments. These findings provide quantitative evidence of a pervasive recency bias in LLMs and highlight the importance of effective bias-mitigation strategies.

cs.IR↗
arXiv2026

UniRec: Cross-stage Multi-Task Fusion with Preference Alignment for Cascaded Recommender Systems

Industrial recommender systems cascade stages with different objectives, feature spaces, and latency constraints. Optimizing pre-ranking and ranking separately induces cross-stage inconsistency: upstream models may filter out items preferred by downstream rankers, while independently tuned downstream fusion can offset upstream improvements. Most existing multi-task fusion methods target the ranking stage alone, and cross-stage methods often align with a downstream-derived score, leaving joint optimization of fusion modules across cascaded stages largely unexplored. We propose UniRec, a Unified Cross-stage Recommendation Fusion model. First, the two fusion agents partially share input embeddings in a single computation graph, allowing gradients from either stage to propagate through the shared representations. Second, a dual-axis preference alignment objective coordinates the two stages: horizontally, a compact aggregation term reorganizes dozens of pairwise objectives over heterogeneous prior signals into bidirectional preference evidence; vertically, a cross-stage consistency term transfers downstream pairwise preferences to the upstream fusion score. Third, we introduce attribute group-relative regularization, which computes relative advantages and normalizes policy updates within each attribute group, ensuring that uniformly promoting all items in a high-reward group provides no additional optimization gain. Offline experiments demonstrate UniRec consistently outperforms single-stage fusion and cross-stage coordination baselines; online A/B experiments show a 0.616% gain in app usage duration. UniRec has been fully deployed on the Kuaishou platform.

cs.IR↗
arXiv2026

Scaling Articulated Rationales for MLLM-based Recommendation

We presented SARA, an industrial framework that transforms sparse articulated user rationales into scalable recommendation signals. Its data engine curates questionnaire responses into SARA-HQ, providing explicit preference supervision for aligning SARA-7B through SFT and Quality-Refining DPO. This alignment extends rationale generation from $86{,}564$ questionnaire-covered authors to the full $10$M-author space. SARA-Ranker translates the generated positive and negative rationales into features for user--author interaction modeling and negative-feedback history modeling, connecting articulated reasons to production ranking. Evaluation on unseen authors demonstrates that SARA-7B generates more specific, relevant, and grounded rationales than the evaluated general-purpose MLLMs. On top of a strong industrial ranking baseline with multimodal features, separate online A/B tests show that positive-rationale integration increases watch time by $0.99\%$, while negative-rationale integration reduces Hate feedback by $8.16\%$. Daily refresh and more than $30$ days of production deployment further demonstrate the operational feasibility of the approach. These findings establish articulated rationales as a useful complement to behavioral and content signals, and demonstrate a practical role for MLLMs in scaling sparse human explanations into preference information that improves industrial recommendation.

cs.IR↗