Breaking the Curse of Quality Saturation with User-Centric Ranking
This addresses a key bottleneck in ranking systems for search, ads, and recommendation, but it appears incremental as it builds on existing dyadic interaction data.
The paper tackles the problem of diminishing returns in ranking models for search, ads, and recommendation by identifying item-centric formulation as a root cause and introducing user-centric ranking as an alternative. It shows that this formulation enables training better-converged models on larger datasets, though no concrete numbers are provided.
A key puzzle in search, ads, and recommendation is that the ranking model can only utilize a small portion of the vastly available user interaction data. As a result, increasing data volume, model size, or computation FLOPs will quickly suffer from diminishing returns. We examined this problem and found that one of the root causes may lie in the so-called ``item-centric'' formulation, which has an unbounded vocabulary and thus uncontrolled model complexity. To mitigate quality saturation, we introduce an alternative formulation named ``user-centric ranking'', which is based on a transposed view of the dyadic user-item interaction data. We show that this formulation has a promising scaling property, enabling us to train better-converged models on substantially larger data sets.