Nov 2026· Proceedings of the 35th ACM International Conference on Information and Knowledge Management· pp. 7841-7846· 0 citations· 30 references
TL;DR
Deployed in a large-scale e-commerce platform under strict latency constraints, CAIM yields a 6.9% increase in orders and a 7.7% increase in Gross Merchandise Value, validating its effectiveness and practical applicability.
Abstract
Relevant recommendation is essential for e-commerce conversion and is typically initiated by a user-clicked trigger item. However, user intent is inherently dynamic and multifaceted. Existing methods with static trigger representations cannot adequately capture real-time intent drift or disentangle fine-grained preferences such as brand versus category. We propose Context-Aware User Intent Modeling (CAIM) to address these limitations. CAIM incorporates an Intention Enhancement Module (IEM) that integrates real-time context to dynamically reconstruct the trigger representation, serving as an adaptive anchor. An Intention Retrieval Module (IRM) retrieves relevant long-term user behaviors, while a Sub-intention Extraction Module (SIEM) decomposes the overall intent into attribute-level sub-intentions. Experiments on industrial datasets show that CAIM outperforms state-of-the-art baselines. Deployed in a large-scale e-commerce platform under strict latency constraints, CAIM yields a 6.9% increase in orders and a 7.7% increase in Gross Merchandise Value (GMV), validating its effectiveness and practical applicability.
This work introduces SHOPPER, which addresses bottlenecks in large-scale e-commerce recommendation systems through Semantic history, order-aware local intent modeling, and serving-aware factorization, enabling deep target-aware summarization.
Amit Jaspal, Nicolas Bievre, Ashkan Sadeghi et al.· 0 citations
User behavior sequences are often driven by stable, high-level intents. Existing intent-aware sequential recommendation methods typically extract intents from derived sequence representations to serve as post-hoc supervisory signals. Although effective, this post-hoc paradigm prevents intents from explicitly participat...
Yi-Fan Shao, Pei-Lin Zhou, Shou-Jin Wang et al.· Proceedings of the 35th ACM...· 0 citations
This work proposes an agentic, human-aligned prompting framework that generates multi-category relevance guidelines and distills teacher-model judgments into an efficient student model for scalable evaluation of DPA retrieval systems, enabling faster and more reliable DPA retrieval iteration before A/B testing.
Wen Zhang, Xiao Bai, Hsiang-Wei Chao et al.· Proceedings of the 35th ACM...· 0 citations
Session-based Recommendation (SBR) aims to predict users' next interaction based on short-term anonymous behavior sequences. Yet accurately modeling the dynamic switching and hierarchical dependency of user intent remains a critical challenge. Existing methods typically treat intents as parallel signals thus ignoring t...
Yu Qing, Xiao Wang, Zhe Wang et al.· Proceedings of the 35th ACM...· 0 citations
A personalized quality-aware reranking framework that explicitly incorporates product quality into the ranking objective, and proposes a dual-model architecture: Personalized NQD (PNQD), which estimates user-specific dissatisfaction risk for a user–product pair, and Retention-Driven Alpha (RDA), which adaptively contro...
Siddhant Doshi, Krish Rupapara, Bhavuk Singhal et al.· Proceedings of the 35th ACM...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.