Recommending what video to watch next: a multitask ranking system

By Zhe Zhao, Lichan Hong, Li Wei, Jilin Chen, Aniruddh Nath, Shawn Andrews, Aditee Kumthekar, M. Sathiamoorthy, Xinyang Yi, Ed H. Chi. 2019

418 citations recorded by the source corpus.

In this paper, we introduce a large scale multi-objective ranking system for recommending what video to watch next on an industrial video sharing platform. The system faces many real-world challenges, including the presence of multiple competing ranking objectives, as well as implicit selection biases in user feedback. To tackle these challenges, we explored a variety of soft-parameter sharing techniques such as Multi-gate Mixture-of-Experts so as to efficiently optimize for multiple ranking objectives. Additionally, we mitigated the selection biases by adopting a Wide & Deep framework. We demonstrated that our proposed techniques can lead to substantial improvements on recommendation quality on one of the world's largest video sharing platforms.

Subfields: Computer Vision and Pattern Recognition, Information Systems, Management Science and Operations Research, Recommender Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Information Systems

In Management Science and Operations Research

In Recommender Systems

Citations in the corpus

Pathways: Asynchronous Distributed Dataflow for ML (2022, background): “…neural network training (Jia et al., 2020), neural architecture search (Pham et al., 2018), and multi-modal multi-task learning systems (Ma et al., 2018; Lepikhin et al., 2020; Zhao et al., 2019) are examples of inherently heterogeneous and dynamic tasks that do not ﬁt naturally in the SPMD model.”
DSelect-k: Differentiable Selection in the Mixture of Experts with Applications to Multi-Task Learning (2021, methodology): “Architecture and Dataset: The system consists of a candidate generator followed by a multi-task ranking model, and it adopts a framework similar to [40, 34].”
Variable-Shot Adaptation for Online Meta-Learning (2020, background): “Indeed, the real world is arguably so complex that, for all practical purposes, learning is never truly finished, and any real system in open-world settings will need to improve and finetune perpetually (Chen & Asch, 2017; Zhao et al., 2019).”
Small Towers Make Big Differences (2020, methodology): “For example, an object detection algorithm may involve predicting both type and position of an object (Girshick, 2015); a content recommendation system may care about both short-term conversion rate and longterm retention probabilities of the users (Zhao et al., 2019).”
RecSSD: near data processing for solid state drive based recommendation inference (2021, methodology): “Recommendation algorithms are used across a variety of Internet services such as social media, entertainment, e-commerce, and search [20, 37, 43–45].”
Low-Precision Hardware Architectures Meet Recommendation Model Inference at Scale (2021, background): “Recommender systems have evolved with increasing model complexities: from simple content and collaborative ﬁltering, to matrix factorization [37, 50], multi-layer perceptrons and deep learning models [18, 47, 54, 55, 64, 65, 67].”
Is ChatGPT a Good Recommender? A Preliminary Study (2023, methodology): “As a crucial technique for addressing information overload and enhancing user experience, recommendation systems have witnessed significant advancements over the past decade and have been widely used in various web applications such as product recommendation [32, 49, 51, 59], video recommendation [3”
RecNMP: Accelerating Personalized Recommendation with Near-Memory Processing (2019, background): “sparse features are broadly observable across many alternative recommendation models [1], [25], [49], [52], [63], [83].”
RecShard: statistical feature-based memory optimization for industry-scale neural recommendation (2022, background): “A significant fraction of deep learning compute cycles in industry-scale data centers can be attributed to deep learning recommendation models (DL-RMs) [3, 4, 9, 11, 20, 34, 43, 47, 48].”
The Benchmark Lottery (2021, background): “( i ) Different recommendation platforms tackle slightly different problems (e.g retrieval [Yi et al., 2019], ranking ([Pei et al., 2019]), or multitask learning ([Zhao et al., 2019])), and each requires their own evaluation setup.”
DeepRecSys: A System for Optimizing End-To-End At-Scale Neural Recommendation Inference (2020, methodology): “As their sophistication increases with more and better quality data, recommendation algorithms have evolved from simple rule-based or nearest neighbor-based designs [1] to deep learning approaches [2]–[7].”