TwERC: High Performance Ensembled Candidate Generation for Ads Recommendation at Twitter
This work addresses the computational efficiency and quality trade-offs in large-scale ads recommendation for social media platforms, presenting an incremental improvement with validated gains.
The paper tackles the candidate generation phase in Twitter's ads recommendation system by introducing TwERC, a heterogeneous architecture combining a real-time light ranker with two sourcing strategies, resulting in revenue gains of 4.08% from a graph-based strategy and 1.38% from a rankscore-based strategy.
Recommendation systems are a core feature of social media companies with their uses including recommending organic and promoted contents. Many modern recommendation systems are split into multiple stages - candidate generation and heavy ranking - to balance computational cost against recommendation quality. We focus on the candidate generation phase of a large-scale ads recommendation problem in this paper, and present a machine learning first heterogeneous re-architecture of this stage which we term TwERC. We show that a system that combines a real-time light ranker with sourcing strategies capable of capturing additional information provides validated gains. We present two strategies. The first strategy uses a notion of similarity in the interaction graph, while the second strategy caches previous scores from the ranking stage. The graph based strategy achieves a 4.08% revenue gain and the rankscore based strategy achieves a 1.38% gain. These two strategies have biases that complement both the light ranker and one another. Finally, we describe a set of metrics that we believe are valuable as a means of understanding the complex product trade offs inherent in industrial candidate generation systems.