[论文解读] Perceive Your Users in Depth: Learning Universal User Representations from Multiple E-commerce Tasks

Yabo Ni, Dan Ou+5 more|arXiv (Cornell University)|May 28, 2018

Recommender Systems and Techniques参考文献 35被引用 25

一句话总结

本文提出深度用户感知网络（DUPN），一种多任务学习框架，利用LSTM和注意力机制，从跨多个任务的异构电商行为序列中学习通用用户表征。通过在搜索、推荐及其他任务之间端到端共享表征，DUPN提升了个性化推荐的准确性，并实现了向新任务的高效迁移，在淘宝的在线A/B测试中实现了2.23%的CTR提升和3.17%的销售额增长。

ABSTRACT

Tasks such as search and recommendation have become increas- ingly important for E-commerce to deal with the information over- load problem. To meet the diverse needs of di erent users, person- alization plays an important role. In many large portals such as Taobao and Amazon, there are a bunch of di erent types of search and recommendation tasks operating simultaneously for person- alization. However, most of current techniques address each task separately. This is suboptimal as no information about users shared across di erent tasks. In this work, we propose to learn universal user representations across multiple tasks for more e ective personalization. In partic- ular, user behavior sequences (e.g., click, bookmark or purchase of products) are modeled by LSTM and attention mechanism by integrating all the corresponding content, behavior and temporal information. User representations are shared and learned in an end-to-end setting across multiple tasks. Bene ting from better information utilization of multiple tasks, the user representations are more e ective to re ect their interests and are more general to be transferred to new tasks. We refer this work as Deep User Perception Network (DUPN) and conduct an extensive set of o ine and online experiments. Across all tested ve di erent tasks, our DUPN consistently achieves better results by giving more e ective user representations. Moreover, we deploy DUPN in large scale operational tasks in Taobao. Detailed implementations, e.g., incre- mental model updating, are also provided to address the practical issues for the real world applications.

研究动机与目标

• 通过在多个搜索与推荐任务之间共享用户表征知识，解决电商业务中个性化推荐效果不佳的问题。
• 学习能够反映深层用户兴趣的通用、可泛化的用户表征，基于多样化的行为序列。
• 通过训练单一统一网络而非每个任务单独建模，提升模型效率与可迁移性。
• 实现在淘宝搜索引擎等大规模生产系统中的实时、可扩展推理。
• 为实际运营中的深度学习系统提供增量训练与模型部署的实用指导。

提出的方法

• 利用双向LSTM对用户行为序列（点击、购买、收藏等）进行编码，结合时间与内容特征。
• 采用一种新颖的注意力机制，根据内容、行为类型和时间动态分配序列中各项的权重。
• 在端到端的多任务学习框架中，跨多个任务（如CTR预测、排序、价格偏好）共享用户表征。
• 引入网络解耦技术，将用户表征学习与项目特定评分解耦，实现高效的在线推理。
• 通过每日微调实现增量模型更新，无需全量重训即可适应用户偏好的动态变化。
• 使用包含2000个worker的分布式TensorFlow系统，实现可扩展的训练与在线部署。

实验结果

研究问题

• RQ1 与任务特定模型相比，跨多个电商业务任务学习到的共享用户表征是否能提升个性化推荐性能？
• RQ2 注意力机制在捕捉异构行为序列中用户动态兴趣方面的有效性如何？
• RQ3 通用用户表征在多大程度上可迁移至新出现的、未见过的任务？
• RQ4 哪些实用策略能够实现大规模深度学习模型在生产电商系统中的高效部署与实时推理？
• RQ5 多任务学习在多大程度上增强了用户表征的鲁棒性与泛化能力？

主要发现

• 在淘宝的在线A/B测试中，DUPN使点击率（CTR）提升了2.23%，七天内销售额增长3.17%。
• 价格偏好分类的精确率从33.2%提升至44.2%，所有价格层级的召回率均提升2%–10%。
• 价格偏好任务的整体精确率由33.2%提升至44.2%，其中在最便宜和最贵的商品类别中提升最为显著。
• 增量微调将训练时间从3–4天缩短至10小时以内，同时保持模型精度与适应性。
• 通过网络解耦，实现了高效的在线推理：每个查询仅需一次计算用户表征，并可在数千个商品间复用。
• 模型展现出强大的可迁移性，在无需任务特定微调的情况下，即在新任务（价格偏好）上达到高性能。