Practical Lessons on Optimizing Sponsored Products in eCommerce
This work addresses incremental improvements in ad system optimization for e-commerce platforms, focusing on practical solutions without requiring structural changes to existing models.
The paper tackled multiple sponsored product optimization problems in e-commerce ad systems, such as position-based de-biasing and click-conversion multi-task learning, by proposing a practical machine learning framework that improved multiple evaluation metrics on real-world traffic logs.
In this paper, we study multiple problems from sponsored product optimization in ad system, including position-based de-biasing, click-conversion multi-task learning, and calibration on predicted click-through-rate (pCTR). We propose a practical machine learning framework that provides the solutions to such problems without structural change to existing machine learning models, thus can be combined with most machine learning models including shallow models (e.g. gradient boosting decision trees, support vector machines). In this paper, we first propose data and feature engineering techniques to handle the aforementioned problems in ad system; after that, we evaluate the benefit of our practical framework on real-world data sets from our traffic logs from online shopping site. We show that our proposed practical framework with data and feature engineering can also handle the perennial problems in ad systems and bring increments to multiple evaluation metrics.