Meta Lattice: Model Space Redesign for Cost-Effective Industry-Scale Ads Recommendations

2025-12-09

AI Summary

Addressing the dual challenges of cross-domain data fragmentation and high inference cost in industrial-scale advertising recommendation, this paper proposes a model–data–system co-optimization paradigm. Methodologically, it pioneers the extension of multi-domain, multi-objective learning to the model architecture design level, introducing a unified model that enables cross-domain knowledge sharing and data fusion; it further integrates knowledge distillation with distributed inference optimization to achieve end-to-end model compression and deployment. The approach significantly improves both recommendation quality and system efficiency: real-world deployment yields a 10% revenue increase, an 11.5% rise in user satisfaction, a 6% uplift in conversion rate, and a 20% reduction in computational resource consumption. The core contribution lies in the novel unification of modeling space, data governance, and system-level optimization—establishing a scalable, cost-effective framework for large-scale, multi-scenario recommendation systems.

Technology Category

• Data Mining & Knowledge Management: Recommender Systems
• Search and Optimization: Learning to Search
• Machine Learning: Learning on the Edge & Model Compression

Application Category

• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

The rapidly evolving landscape of products, surfaces, policies, and regulations poses significant challenges for deploying state-of-the-art recommendation models at industry scale, primarily due to data fragmentation across domains and escalating infrastructure costs that hinder sustained quality improvements. To address this challenge, we propose Lattice, a recommendation framework centered around model space redesign that extends Multi-Domain, Multi-Objective (MDMO) learning beyond models and learning objectives. Lattice addresses these challenges through a comprehensive model space redesign that combines cross-domain knowledge sharing, data consolidation, model unification, distillation, and system optimizations to achieve significant improvements in both quality and cost-efficiency. Our deployment of Lattice at Meta has resulted in 10% revenue-driving top-line metrics gain, 11.5% user satisfaction improvement, 6% boost in conversion rate, with 20% capacity saving.

Problem

• Addresses data fragmentation and high costs in large-scale ad recommendations
• Proposes a framework for cross-domain knowledge sharing and model unification
• Aims to improve recommendation quality while reducing infrastructure expenses

Innovation

• Model space redesign for cross-domain knowledge sharing
• Combines data consolidation, model unification, and distillation
• System optimizations for quality and cost-efficiency improvements