GAP-Net: Calibrating User Intent via Gated Adaptive Progressive Learning for CTR Prediction

TL;DR

GAP-Net introduces a Triple Gating framework for CTR prediction to address three fundamental bottlenecks: the Attention Sink, Static Intent, and Rigid View Aggregation. It combines Adaptive Sparse-Gated Attention (ASGA) at the micro level, Gated Cascading Query Calibration (GCQC) at the meso level, and Context-Gated Denoising Fusion (CGDF) at the macro level to progressively denoise features, evolve user intent with real-time context, and adaptively fuse multi-view sequences. The approach yields state-of-the-art results on industrial datasets (XMart and KuaiVideo) and gains in online A/B tests (GMV, CVR, V2P), demonstrating robustness to interaction noise and intent drift. Overall, GAP-Net shows that systematic gating across scales improves both the quality and stability of recommendations in dynamic user environments.

Abstract

Sequential user behavior modeling is pivotal for Click-Through Rate (CTR) prediction yet is hindered by three intrinsic bottlenecks: (1) the "Attention Sink" phenomenon, where standard Softmax compels the model to allocate probability mass to noisy behaviors; (2) the Static Query Assumption, which overlooks dynamic shifts in user intent driven by real-time contexts; and (3) Rigid View Aggregation, which fails to adaptively weight heterogeneous temporal signals according to the decision context. To bridge these gaps, we propose GAP-Net (Gated Adaptive Progressive Network), a unified framework establishing a "Triple Gating" architecture to progressively refine information from micro-level features to macro-level views. GAP-Net operates through three integrated mechanisms: (1) Adaptive Sparse-Gated Attention (ASGA) employs micro-level gating to enforce sparsity, effectively suppressing massive noise activations; (2) Gated Cascading Query Calibration (GCQC) dynamically aligns user intent by bridging real-time triggers and long-term memories via a meso-level cascading channel; and (3) Context-Gated Denoising Fusion (CGDF) performs macro-level modulation to orchestrate the aggregation of multi-view sequences. Extensive experiments on industrial datasets demonstrate that GAP-Net achieves substantial improvements over state-of-the-art baselines, exhibiting superior robustness against interaction noise and intent drift.

Figures (3)

• Figure 1: Illustration of intrinsic blind spots. (a) Micro-Level: Inductive Bias Flaw. The standard Softmax enforces a strict sum-to-one constraint, forcing the model to assign spurious weights to irrelevant noise. (b) Meso-Level: Representation Gap. Static target embeddings fail to capture dynamic intent shifts driven by real-time context cues (e.g., shifting from "daily meal" to "social dining"), leading to intent misalignment. (c) Macro-Level: Rigid Fusion. Static aggregation lacks adaptivity, failing to dynamically modulate the trade-off between short-term impulses and long-term habits based on the specific decision scenario.
• Figure 2: An Overview of the proposed GAP-Net. The framework employs a "Triple Gating" architecture for progressive denoising and calibration: (a) Micro-Level (ASGA): Replaces Softmax with learnable sparse gating, eliminating the strict sum-to-one constraint. (b) Meso-Level (GCQC): Evolves static target embeddings by fusing real-time context triggers. (c) Macro-Level (CGDF): A context-aware network that dynamically modulates fusion weights for heterogeneous views.
• Figure 3: Impact of Gating Strategy in CGDF