Learning Graph Meta Embeddings for Cold-Start Ads in Click-Through Rate Prediction

TL;DR

This paper introduces Graph Meta Embedding models that leverage graph neural networks and meta learning to generate effective initial embeddings for new ads, significantly improving CTR prediction in cold-start scenarios.

Contribution

The paper proposes novel GME models that incorporate information from both new ads and existing old ads, enhancing cold-start ad embedding initialization.

Findings

GME models outperform existing CTR prediction models in cold-start scenarios.

GME models improve performance in warm-up scenarios with limited training data.

The approach is applicable to both CTR and CVR prediction tasks.

Abstract

Click-through rate (CTR) prediction is one of the most central tasks in online advertising systems. Recent deep learning-based models that exploit feature embedding and high-order data nonlinearity have shown dramatic successes in CTR prediction. However, these models work poorly on cold-start ads with new IDs, whose embeddings are not well learned yet. In this paper, we propose Graph Meta Embedding (GME) models that can rapidly learn how to generate desirable initial embeddings for new ad IDs based on graph neural networks and meta learning. Previous works address this problem from the new ad itself, but ignore possibly useful information contained in existing old ads. In contrast, GMEs simultaneously consider two information sources: the new ad and existing old ads. For the new ad, GMEs exploit its associated attributes. For existing old ads, GMEs first build a graph to connect them with new ads, and then adaptively distill useful information. We propose three specific GMEs from different perspectives to explore what kind of information to use and how to distill information. In particular, GME-P uses Pre-trained neighbor ID embeddings, GME-G uses Generated neighbor ID embeddings and GME-A uses neighbor Attributes. Experimental results on three real-world datasets show that GMEs can significantly improve the prediction performance in both cold-start (i.e., no training data is available) and warm-up (i.e., a small number of training samples are collected) scenarios over five major deep learning-based CTR prediction models. GMEs can be applied to conversion rate (CVR) prediction as well.