超越同购关系：Allegro互补推荐系统的演进

文章背景与核心概要

在电商平台上，当客户购买一台专业相机时，标准推荐模型往往难以区分用户究竟需要兼容的镜头、通用的三脚架，还是仅仅是另一台相机机身。为了解决这一对构建完整购物篮至关重要的互补商品推荐（Complementator Product Recommendation）难题，波兰电商巨头 Allegro 的研究人员推出了 AlleCompanion。

该系统已在 Allegro.com 上实现生产规模部署，它通过结合数据级过滤启发式算法与受类别约束的双塔架构（由类别适配器引导），成功将充满噪声的行为信号转化为精准的语义兼容性。此外，作者引入了 ComCat（多源互补类别映射层），通过融合专家规则、人机回环反馈、基于大语言模型（LLM）的推理以及统计挖掘，从海量流量中提炼出有意义的模式。该框架服务于每月超过 2000 万的活跃用户，有效提升了自然发现的GMV（商品交易总额），并显著推动了赞助广告位的收入增长。

文档详情 (Document Details)

摘要 (Abstract)

互补商品推荐对于构建完整的购物篮至关重要，然而标准模型往往无法区分仅仅是“一起购买”的商品与真正“协同使用”的商品。在本文中，我们提出了 AlleCompanion：这是一个部署在 Allegro.com 上的生产级检索框架，它将充满噪声的行为信号转化为精确的语义兼容性。

Complementary Product Recommendation is vital for comprehensive basket building, yet standard models often fail to distinguish between items that are merely bought together and those that truly work together. In this paper, we present AlleCompanion: a production-scale retrieval framework deployed at Allegro.com that transforms noisy behavioural signals into precise semantic compatibility.

我们通过将数据级过滤启发式算法与类别约束的双塔架构相结合，缓解了大尺度同购流量中固有的噪声。在该框架内，类别适配器引导嵌入空间中的模型，将候选商品限制在逻辑互补的边界内。由于在大规模下建模真实的真实用户行为本质上非常困难，我们引入了 ComCat（多源互补类别映射）。ComCat 充当一个转换层，将来自噪声流量的有意义模式提炼为一个可维护且可控的解决方案，它整合了专家规则、人机回环反馈、基于 LLM 的推理以及统计挖掘。

We mitigate the intrinsic noise in large-scale co-purchase traffic by combining data-level filtering heuristics with a category-constrained Two Tower architecture. Within this framework, the Category Adapter guides the model in the embedding space, constraining candidates within logically complementary boundaries. Since modelling authentic user behaviour at scale is inherently difficult, we introduce ComCat, a multi-source Complementary Categories Mapping. ComCat acts as a translational layer that distils meaningful patterns from noisy traffic into a maintainable and controllable solution, integrating expert rules, human-in-the-loop feedback, LLM-based reasoning, and statistical mining.

我们的实验结果表明，将显式的类别级约束与神经架构相结合，可以有效过滤同购噪声，从而呈现满足真实世界用户需求的推荐结果。该框架每月为超过 2000 万活跃用户提供服务，为自然发现带来了显著的归因 GMV 提升，并推动了赞助广告位的实质性收入增长。

Our experimental results demonstrate that combining explicit category-level constraints with neural architectures effectively filters out co-purchase noise to surface recommendations that satisfy real-world user needs. Serving over 20 million active users monthly, the framework delivers significant uplifts in attributed GMV for organic discovery and drives substantial revenue growth in sponsored placements.