Trinity: A Scenario-Aware Recommendation Framework for Large-Scale Cold-Start Users

TL;DR

Trinity is a comprehensive recommendation framework that effectively addresses the cold-start problem for large-scale new users by integrating feature engineering, model architecture, and stable updates, demonstrated on a billion-user platform.

Contribution

We introduce Trinity, a novel framework that combines multiple strategies to improve recommendations for new users in new scenarios, surpassing prior model-only approaches.

Findings

Significant offline performance gains in cold-start scenarios.

Notable online engagement improvements in a large-scale deployment.

Effective handling of sparse signals and low-engagement cohorts.

Abstract

Early-stage users in a new scenario intensify cold-start challenges, yet prior works often address only parts of the problem through model architecture. Launching a new user experience to replace an established product involves sparse behavioral signals, low-engagement cohorts, and unstable model performance. We argue that effective recommendations require the synergistic integration of feature engineering, model architecture, and stable model updating. We propose Trinity, a framework embodying this principle. Trinity extracts valuable information from existing scenarios while ensuring predictive effectiveness and accuracy in the new scenario. In this paper, we showcase Trinity applied to a billion-user Microsoft product transition. Both offline and online experiments demonstrate that our framework achieves substantial improvements in addressing the combined challenge of new users in new scenarios.