Time-Aware Prospective Modeling of Users for Online Display Advertising
This work addresses the problem of improving conversion prediction for advertising platforms, though it appears incremental as it builds on existing modeling efforts.
The paper tackles the challenge of predicting user conversions in online display advertising by proposing a time-aware approach to model user activity sequences and capture implicit signals of conversion intent, showing that it outperforms previous methods on two real-world datasets.
Prospective display advertising poses a great challenge for large advertising platforms as the strongest predictive signals of users are not eligible to be used in the conversion prediction systems. To that end efforts are made to collect as much information as possible about each user from various data sources and to design powerful models that can capture weaker signals ultimately obtaining good quality of conversion prediction probability estimates. In this study we propose a novel time-aware approach to model heterogeneous sequences of users' activities and capture implicit signals of users' conversion intents. On two real-world datasets we show that our approach outperforms other, previously proposed approaches, while providing interpretability of signal impact to conversion probability.