Papers

COTTI 推荐框架:对比训练与时间感知推理应对促销漂移

COTTI: Contrastive Training and Time-Aware Inference for Robust Industrial Recommendation under Promotional Drift

Why we picked it

做推荐系统的同学值得看看这篇飞猪的实战论文。它把促销期行为漂移拆成表示学习和分数调整两步,线上十天转化率涨了 3.14%,还给出延迟和召回下降的取舍分析,落地细节比较扎实。

阿里巴巴飞猪团队发布 COTTI 推荐框架,用于解决促销活动导致的用户行为漂移问题。该框架在飞猪数据上取得 0.8127 的离线 AUC,高于 TEN 基线的 0.8101。十天线上实验中,COTTI 相对服务中的 HDR 模型实现转化率提升 3.14%、千次展示收入提升 2.68%。

Body

技术上,COTTI 将历史表示学习与事件时点分数调整分离为两个阶段。跨周期对比对齐(CPC)通过自适应一致性权重正则化同一用户的跨期行为对;主动重加权(RW)依据历史兴趣距离控制促销样本的影响;自适应边界对齐(ABA)结合随机预测波动与时间事件核来调整 logits,预测器参数保持冻结,为 p99 延迟仅增加 2.2ms。

端到端测量显示 p99 延迟为 38.9ms、吞吐量达 14,700 请求/秒。跨品类分析揭示了一个权衡:整体判别能力提升,但新品的付费商品召回下降,这是工业推荐系统在促销场景下需要面对的取舍。

Original · arXiv cs.AI

COTTI: Contrastive Training and Time-Aware Inference for Robust Industrial Recommendation under Promotional Drift

Promotional events shift traffic composition, purchase behavior, and the scale of conversion probabilities on different time horizons. We present \textbf{COTTI} (\textbf{CO}ntrastive \textbf{T}raining and \textbf{T}ime-aware \textbf{I}nference), an industrial recommendation framework that separates historical representation learning from event-time score adjustment. Cross-Period Contrastive Alignment (CPC) regularizes paired histories of the same user with adaptive consistency weights. Active Re-weighting (RW) controls the influence of promotional examples through historical-interest distance. Adaptive Boundary Alignment (ABA) combines stochastic prediction variability with a temporal event kernel to adjust logits while keeping predictor parameters fixed. We analyze the weighted objective and bound ABA-induced probability and ranking changes. On Fliggy data, COTTI attains an offline AUC of 0.8127, compared with 0.8101 for TEN and 0.8123 for TEN with ABA. Cross-category analysis reveals improved discrimination alongside declines in paid-item recall for novel categories. End-to-end measurements give a p99 latency of 38.9\,ms and throughput of 14,700 requests/s; enabling ABA adds 2.2\,ms to p99. A ten-day online experiment against the serving HDR model yields relative gains of 3.14\% in conversion rate and 2.68\% in revenue per mille. Together, these results characterize the trade-off between aggregate prediction, novel-category discovery, and serving cost in promotional recommendation.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include recommendation systems, CPC, contrastive learning.

Source attribution: curated from arXiv cs.AI (@Maolei Huang, Jing Tian, Dexuan Duan, Fangfang Lin, Huawei Cao, Shuhan Song).