PRISM: Refracting the Entangled User Behavior Space for E-Commerce Search

2026-05-08

AI Summary

In e-commerce search, user behavior signals suffer from entangled drift due to exposure bias, feedback loops, and coupling with semantic matching, leading to confusion between preference estimation and relevance modeling. To address this, this work proposes PRISM, a novel framework that explicitly models the dynamic interaction between user preferences and item relevance for the first time. PRISM introduces three core mechanisms: preference correction to mitigate bias, large language model–driven semantic anchoring to align semantic spaces, and preference-conditioned evidence routing for context-aware adaptive fusion of multi-source behavioral signals. Experiments on two public e-commerce datasets demonstrate that PRISM significantly outperforms strong baselines, effectively enhancing the robustness and semantic consistency of user behavior modeling.

Abstract

E-commerce search systems rely on modeling user behavior to estimate item relevance and user preference, which are typically assumed to be stable and independently learnable signals. However, in practice, user interactions are jointly shaped by exposure mechanisms, feedback loops, and semantic matching, leading to entangled and dynamically drifting behavioral signals. As a result, both preference estimation and relevance modeling suffer from confounding effects and semantic misalignment, which limits the robustness of downstream ranking models. To address this issue, we propose PRISM, a Preference-Relevance Interaction Semantic Modeling framework for e-commerce search behavior prediction. PRISM explicitly models the interaction between user preference and item relevance rather than treating them as independent components. Specifically, it introduces a preference rectification module to iteratively refine user preference under relevance-aware constraints, improving robustness against behavioral confounding. To ensure semantic consistency, we further incorporate a large language model (LLM)-driven semantic anchoring mechanism that leverages positive and negative prototypes to calibrate relevance representations. Finally, a preference-conditioned evidence routing module adaptively aggregates multi-source behavioral signals, enabling context-aware and preference-aligned relevance estimation. Extensive experiments on two public e-commerce benchmarks demonstrate that PRISM consistently outperforms strong baselines, validating the effectiveness of explicitly modeling preference-relevance interaction for robust and semantically grounded search behavior modeling.

Problem

• user behavior entanglement
• preference-relevance confounding
• semantic misalignment
• e-commerce search
• behavioral signal drift

Innovation

• preference-relevance interaction
• semantic anchoring
• preference rectification
• evidence routing
• large language model