Leaving No One Behind: A Multi-Scenario Multi-Task Meta Learning Approach for Advertiser Modeling

Abstract

Advertisers play an essential role in many e-commerce platforms like Taobao and Amazon. Fulfilling their marketing needs and supporting their business growth is critical to the long-term prosperity of platform economies. However, compared with extensive studies on user modeling such as click-through rate predictions, much less attention has been drawn to advertisers, especially in terms of understanding their diverse demands and performance. Different from user modeling, advertiser modeling generally involves many kinds of tasks (e.g. predictions of advertisers' expenditure, active-rate, or total impressions of promoted products). In addition, major e-commerce platforms often provide multiple marketing scenarios (e.g. Sponsored Search, Display Ads, Live Streaming Ads) while advertisers' behavior tend to be dispersed among many of them. This raises the necessity of multi-task and multi-scenario consideration in comprehensive advertiser modeling, which faces the following challenges: First, one model per scenario or per task simply doesn't scale; Second, it is particularly hard to model new or minor scenarios with limited data samples; Third, inter-scenario correlations are complicated, and may vary given different tasks.To tackle these challenges, we propose a multi-scenario multitask meta learning approach (M2M) which simultaneously predicts multiple tasks in multiple advertising scenarios. Specifically, we introduce a novel meta unit that incorporates rich scenario knowledge to learn explicit inter-scenario correlations and can easily scale to new scenarios. Furthermore, we present a meta attention module to capture diverse inter-scenario correlations given different tasks, and a meta tower module to enhance the capability of capturing the representation of scenario-specific features. Compelling results from both offline evaluation and online A/B tests demonstrate the superiority of M2M over state-of-the-art methods.

Geography of Technology Transfer in China

“…First, we choose methods focusing on behavior sequence modeling, including YouTubeNet [6], DIN [44], DIEN [43], BST [3], ATRank [42], DeepFeed [36], DMT [10] which extract user's interest vector from a single type of behavior sequence or multiple types of behavior sequences. Second, we compare with existing MTL methods in RS, including MMoE [20], PLE [25], M2M [38], AITM [35]. Third, Samll-Heads [29] which concentrates on improving the generalization of the task-shared bottom representation.…”

Multi-Task Deep Recommender Systems: A Survey

“…There is a new trend to tackle multi-task and multiscenario problems in a unified model in an end-to-end manner. For example, M2M [78] proposes to use meta learning to extract specific information from scenario knowledge as dynamic weights considering inter-scenario correlations. AESM 2 [87] proposes to adaptively select shared and specific experts by calculating relevance score via gating mechanism, and stack multiple layers to model hierarchical scenario structure.…”

Multi-Scenario Ranking with Adaptive Feature Learning

“…In this way, it effectively alleviates the noise caused by other scenarios and improves the effectiveness of feature extraction. Besides, Zhang et al [25] utilize expert networks to solve multi-scenario and multi-task problems on the advertiser's side and use the meta network to express the scenario information explicitly. Zou et al [29] propose a novel expert network structure with automatic selection of fine granularity.…”