HatLLM uses hierarchical attention masks to strengthen cross-item collaboration, achieving a 9.13% average gain.

Yu Cui, Feng Liu, Jiawei Chen et al.

Proposes Hierarchical Semantic RL (HSRL) using fixed semantic space to handle large, dynamic action spaces in recommendation systems.

Minmao Wang, Xingchen Liu, Shijie Yi et al.

Proposed ISMIE framework models information seeking via components, variables, activities; validated in misinformation and AI content trust scenarios.

Shuoqi Sun, Danula Hettiachchi, Damiano Spina

ModernVBERT is a 250M-parameter vision-language encoder that outperforms larger models in document retrieval by optimizing attention masks, image resolution, and training strategies.

Paul Teiletche, Quentin Macé, Max Conti et al.

Survey of industrial recommender systems highlighting challenges in data scale and real-time requirements.

Kuan Zou, Aixin Sun

MME-SID combines multimodal quantization and semantic-ID initialization; on Amazon Beauty, tau rises from 0.0550 to 0.3714.

Yuhao Wang, Junwei Pan, Xinhang Li et al.

Defines and quantifies algorithm adaptation bias in online recommender system experiments, validated through real-world A/B tests, emphasizing ecosystem feedback effects.

Chen Zheng, Zhenyu Zhao

OneRec-V2 introduces a lazy decoder architecture, reducing 94% computation, scaling to 8B parameters, and improving real-world user engagement.

Guorui Zhou, Hengrui Hu, Hongtao Cheng et al.

MPFormer employs a target-conditioned Transformer with dynamic quota adjustment for multi-objective personalized retrieval, boosting coverage and efficiency.

Yijia Sun, Shanshan Huang, Linxiao Che et al.

Proposes HLLM-Creator, a hierarchical LLM framework for efficient personalized content generation, achieving 0.476% CTR increase in industrial ad scenarios.

Junyi Chen, Lu Chi, Siliang Xu et al.

PREMIR framework leverages cross-modal question generation with large language models to achieve state-of-the-art zero-shot multimodal document retrieval, outperforming baselines.

Yejin Choi, Jaewoo Park, Janghan Yoon et al.

OPERA combines reinforcement learning with a hierarchical planner-executor architecture, achieving 57.3% EM on HotpotQA and superior multi-hop reasoning.

Yu Liu, Yanbing Liu, Fangfang Yuan et al.

ProxyRL-FRS combines ProxyNCF and reinforcement learning for efficient client selection, boosting federated recommendation accuracy.

Liang Qu, Jianxin Li, Wei Yuan et al.

Introduces KuaiLive, a large-scale, real-time, multi-behavior dataset from Kuaishou for live streaming recommendation, supporting multi-task learning.

Changle Qu, Sunhao Dai, Ke Guo et al.

CFRR employs inverse propensity scoring with self-normalization to correct bias in user-to-user matching, improving ranking accuracy and fairness.

Kazuki Kawamura, Takuma Udagawa, Kei Tateno

Proposes a data-driven multi-task learning (MTL) framework that enhances rare conversion prediction (<1%) in online advertising, achieving 0.69% AUC lift offline and 2% CPA reduction online.

Yuval Dishi, Ophir Friedler, Yonatan Karni et al.

Proposes utility-based passage selection via distillation, reducing large model costs while improving complex QA performance.

Hengran Zhang, Keping Bi, Jiafeng Guo et al.

VERIRAG employs structured analysis with Veritable taxonomy and Toulmin model to detect methodological flaws, improving F1 by at least 19 points.

Shubham Mohole, Hongjun Choi, Shusen Liu et al.

RankMixer employs multi-head token mixing and sparse MoE, scaling parameters by 100x, boosting MFU from 4.5% to 45%, demonstrating superior industrial recommendation performance.

Jie Zhu, Zhifang Fan, Xiaoxie Zhu et al.

SPAR employs multi-agent architecture with RefChain and query evolution, achieving up to +56% F1 on AutoScholar.

Xiaofeng Shi, Yuduo Li, Qian Kou et al.