PCR-CA: Parallel Codebook Representations with Contrastive Alignment for Multiple-Category App Recommendation

2025-08-25

AI Summary

To address insufficient personalization in recommendation systems caused by semantic overlap among multi-category applications, this paper proposes PCR-CA, an end-to-end CTR prediction framework. Its core contributions are threefold: (1) a parallel codebook vector quantized autoencoder (VQ-AE) that discretizes and disentangles multi-dimensional application semantics; (2) user-level and item-level contrastive alignment losses that explicitly integrate semantic representations with collaborative signals; and (3) a dual-attention mechanism jointly modeling ID-based and semantic features. Extensive experiments on large-scale industrial datasets demonstrate improvements of +0.76% in AUC overall and +2.15% for long-tail applications. Online A/B testing shows significant gains of +10.52% in CTR and +16.30% in CVR. The framework has been fully deployed in the Microsoft App Store.

Technology Category

• Search and Optimization: Applications
• Cognitive Modeling & Cognitive Systems: Applications
• Computer Vision: Applications

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Applications of semantic technologies for improving search, browsing, recommendation, personalization

Abstract

Modern app store recommender systems struggle with multiple-category apps, as traditional taxonomies fail to capture overlapping semantics, leading to suboptimal personalization. We propose PCR-CA (Parallel Codebook Representations with Contrastive Alignment), an end-to-end framework for improved CTR prediction. PCR-CA first extracts compact multimodal embeddings from app text, then introduces a Parallel Codebook VQ-AE module that learns discrete semantic representations across multiple codebooks in parallel -- unlike hierarchical residual quantization (RQ-VAE). This design enables independent encoding of diverse aspects (e.g., gameplay, art style), better modeling multiple-category semantics. To bridge semantic and collaborative signals, we employ a contrastive alignment loss at both the user and item levels, enhancing representation learning for long-tail items. Additionally, a dual-attention fusion mechanism combines ID-based and semantic features to capture user interests, especially for long-tail apps. Experiments on a large-scale dataset show PCR-CA achieves a +0.76% AUC improvement over strong baselines, with +2.15% AUC gains for long-tail apps. Online A/B testing further validates our approach, showing a +10.52% lift in CTR and a +16.30% improvement in CVR, demonstrating PCR-CA's effectiveness in real-world deployment. The new framework has now been fully deployed on the Microsoft Store.

Problem

• Improving CTR prediction for multiple-category app recommendations
• Addressing overlapping semantics in traditional app taxonomies
• Enhancing representation learning for long-tail items

Innovation

• Parallel Codebook VQ-AE for discrete semantic representations
• Contrastive alignment loss bridges semantic and collaborative signals
• Dual-attention fusion combines ID-based and semantic features