HELIX: Purified and Unified - Rethinking Feature Interaction and Sequence Modeling for Large-Scale Recommendation

2026-09-29

AI Summary

This study addresses the performance bottlenecks and scaling law limitations arising from the independent scaling of feature interaction and sequence modeling in industrial recommender systems. To this end, we propose HELIX, an architecture that introduces a novel purified unification paradigm. By interleaving sequential retrieval with feature interaction and enforcing unidirectional information flow, HELIX enables efficient joint scaling along both axes. Furthermore, it incorporates a hybrid tokenization mechanism and amortized sequence state reuse to substantially reduce computational overhead while preserving cross-depth communication, thereby supporting flexible asymmetric scaling. Experimental results demonstrate significant improvements in offline metrics. Notably, online A/B testing reveals an approximate 6% per-user increase in Gross Merchandise Value (GMV) for TikTok e-commerce videos, validating the practical efficacy of the proposed approach in large-scale deployment.

Abstract

Industrial recommendation ranking models typically scale along two modeling axes: feature interaction over heterogeneous user, item, context, and cross features, and sequence modeling over long, informative, and multi-type user behavior histories. We find that scaling either capability in isolation is insufficient, as each exhibits a limited scaling ceiling and a suboptimal scaling-law slope. We conjecture that achieving a more favorable scaling-law slope requires jointly scaling both axes. To support this, we present HELIX, a purified and unified architecture for large-scale recommendation. HELIX interleaves sequence retrieval and feature interaction while enforcing one-way information flow from reusable sequence states to candidate-conditioned mix-tokens. This design preserves cross-depth communication between the two modeling axes while keeping user-side sequence computation amortizable, enabling flexible and asymmetric scaling of sequence modeling and feature interaction. Deployed in TikTok's e-commerce recommendation system, HELIX consistently improves offline CTR AUC, CVR AUC, and other ranking metrics. In online A/B tests, it achieves an approximately 6% increase in e-commerce video GMV per user.

Innovation

• Feature Interaction
• Sequence Modeling
• Scaling Law
• Unified Architecture
• Recommendation System