TWIN V2: Scaling Ultra-Long User Behavior Sequence Modeling for Enhanced CTR Prediction at Kuaishou

doi:doi:10.57702/nmckhotw

TWIN V2: Scaling Ultra-Long User Behavior Sequence Modeling for Enhanced CTR Prediction at Kuaishou

The dataset is a large-scale industrial dataset collected from Kuaishou, containing user interaction data with click-through as labels. The dataset has a maximum history length of 100,000 items for clustering life-cycle user behavior.

BibTex:

@dataset{Zihua_Si_and_Renmin_University_of_China_and_Kuaishou_Technology_Co_Ltd_and_Xiaoxue_Zang_and_Kuaishou_Technology_Co_Ltd_and_Xingchao_Cao_and_Kuaishou_Technology_Co_Ltd_and_Dewei_Leng_and_Kuaishou_Technology_Co_Ltd_and_Yanan_Niu_and_Kuaishou_Technology_Co_Ltd_and_Lin_Guan_and_Kuaishou_Technology_Co_Ltd_and_Jing_Lu_and_Kuaishou_Technology_Co_Ltd_and_Zeyu_Yang_and_Kuaishou_Technology_Co_Ltd_and_Kai_Zheng_and_Kuaishou_Technology_Co_Ltd_and_Yang_Song_and_Kuaishou_Technology_Co_Ltd_and_Zhongxiang_Sun_and_Renmin_University_of_China_and_Yiqun_Hui_and_Kuaishou_Technology_Co_Ltd_and_Yichen_Zheng_and_Kuaishou_Technology_Co_Ltd_and_Chenbin_Zhang_and_Kuaishou_Technology_Co_Ltd_and_Kun_Gai_and_Independent_2025,
    abstract = {The dataset is a large-scale industrial dataset collected from Kuaishou, containing user interaction data with click-through as labels. The dataset has a maximum history length of 100,000 items for clustering life-cycle user behavior.},
    author = {Zihua Si and Renmin University of China and Kuaishou Technology Co., Ltd. and Xiaoxue Zang and Kuaishou Technology Co., Ltd. and Xingchao Cao and Kuaishou Technology Co., Ltd. and Dewei Leng and Kuaishou Technology Co., Ltd. and Yanan Niu and Kuaishou Technology Co., Ltd. and Lin Guan and Kuaishou Technology Co., Ltd. and Jing Lu and Kuaishou Technology Co., Ltd. and Zeyu Yang and Kuaishou Technology Co., Ltd. and Kai Zheng and Kuaishou Technology Co., Ltd. and Yang Song and Kuaishou Technology Co., Ltd. and Zhongxiang Sun and Renmin University of China and Yiqun Hui and Kuaishou Technology Co., Ltd. and Yichen Zheng and Kuaishou Technology Co., Ltd. and Chenbin Zhang and Kuaishou Technology Co., Ltd. and Kun Gai and Independent},
    doi = {10.57702/nmckhotw},
    institution = {No Organization},
    keyword = {'click-through rate', 'recommendation systems', 'user behavior'},
    month = {jan},
    publisher = {TIB},
    title = {TWIN V2: Scaling Ultra-Long User Behavior Sequence Modeling for Enhanced CTR Prediction at Kuaishou},
    url = {https://service.tib.eu/ldmservice/dataset/twin-v2--scaling-ultra-long-user-behavior-sequence-modeling-for-enhanced-ctr-prediction-at-kuaishou},
    year = {2025}
}