Freshness or Accuracy, Why Not Both? Addressing Delayed Feedback via Dynamic Graph Neural Networks

TL;DR

This paper introduces DGDFEM, a dynamic graph neural network approach that effectively addresses delayed feedback in conversion rate prediction, balancing data freshness and label accuracy in online systems.

Contribution

It proposes a novel HLGCN graph convolutional method and a comprehensive pipeline to improve delayed feedback modeling in conversion prediction tasks.

Findings

Outperforms existing methods on industry datasets

Achieves a balance between data freshness and label accuracy

Demonstrates consistent superiority in experiments

Abstract

The delayed feedback problem is one of the most pressing challenges in predicting the conversion rate since users' conversions are always delayed in online commercial systems. Although new data are beneficial for continuous training, without complete feedback information, i.e., conversion labels, training algorithms may suffer from overwhelming fake negatives. Existing methods tend to use multitask learning or design data pipelines to solve the delayed feedback problem. However, these methods have a trade-off between data freshness and label accuracy. In this paper, we propose Delayed Feedback Modeling by Dynamic Graph Neural Network (DGDFEM). It includes three stages, i.e., preparing a data pipeline, building a dynamic graph, and training a CVR prediction model. In the model training, we propose a novel graph convolutional method named HLGCN, which leverages both high-pass and low-pass filters to deal with conversion and non-conversion relationships. The proposed method achieves both data freshness and label accuracy. We conduct extensive experiments on three industry datasets, which validate the consistent superiority of our method.