10 papers
Personalized Task Dependency Graphs for Mitigating Signal Erosion in Multi-Task Recommendation
Fuyuan Liu, Tiandeng Wu, Yaqun Fang +8
Optimizing multiple conversion objectives is a core challenge in industrial recommendation, often limited by signal erosion in rigid architectures. Existing Multi-Task Learning (MT…
Inventory-Grounded Policy-Level Optimization for Training-Free AI Search
Wei Zhou, Tiandeng Wu, Jiandong Ding +2
Early in deployment, an AI search system typically operates over a frequently updated product catalog, so the available items and their properties cannot be treated as stable knowl…
SIDScope: A Diagnostic Resource for Semantic-ID Interfaces in Generative Recommendation
Jiandong Ding, Huijie Qin, Tiandeng Wu +1
Semantic ID tokenizers assign code sequences to catalog items, which generative recommenders predict and resolve back to items. Reusing a released mapping requires checking its cod…
Beyond the Grid: Layout-Informed Multi-Vector Retrieval with Parsed Visual Document Representations
Yibo Yan, Mingdong Ou, Yi Cao +6
Harnessing the full potential of visually-rich documents requires retrieval systems that understand not just text, but intricate layouts, a core challenge in Visual Document Retrie…
Beyond the Flat Sequence: Hierarchical and Preference-Aware Generative Recommendations
Zerui Chen, Heng Chang, Tianying Liu +5
Generative Recommenders (GRs), exemplified by the Hierarchical Sequential Transduction Unit (HSTU), have emerged as a powerful paradigm for modeling long user interaction sequences…
User Long-Term Multi-Interest Retrieval Model for Recommendation
Yue Meng, Cheng Guo, Xiaohui Hu +4
User behavior sequence modeling, which captures user interest from rich historical interactions, is pivotal for industrial recommendation systems. Despite breakthroughs in ranking-…
USD: A User-Intent-Driven Sampling and Dual-Debiasing Framework for Large-Scale Homepage Recommendations
Jiaqi Zheng, Cheng Guo, Yi Cao +3
Large-scale homepage recommendations face critical challenges from pseudo-negative samples caused by exposure bias, where non-clicks may indicate inattention rather than disinteres…
A Generative Re-ranking Model for List-level Multi-objective Optimization at Taobao
Yue Meng, Cheng Guo, Yi Cao +2
E-commerce recommendation systems aim to generate ordered lists of items for customers, optimizing multiple business objectives, such as clicks, conversions and Gross Merchandise V…
ECAT: A Entire space Continual and Adaptive Transfer Learning Framework for Cross-Domain Recommendation
Chaoqun Hou, Yuanhang Zhou, Yi Cao +1
In industrial recommendation systems, there are several mini-apps designed to meet the diverse interests and needs of users. The sample space of them is merely a small subset of th…
Character-based Outfit Generation with Vision-augmented Style Extraction via LLMs
Najmeh Forouzandehmehr, Yijie Cao, Nikhil Thakurdesai +6
The outfit generation problem involves recommending a complete outfit to a user based on their interests. Existing approaches focus on recommending items based on anchor items or s…