Ruiming Tang (Huawei Noah’s Ark), Xiaoqiang Zhu (Mobvista Group), Junfeng Ge (Alibaba), Kuang-chih Lee (AliExpress), Biye Jiang (Alibaba), Xingxing Wang (​​Meituan), Han Zhu (Alibaba), Tao Zhuang (Alibaba), Weiwen Liu (Huawei Noah’s Ark Lab), Kan Ren (Microsoft Research), Weinan Zhang (Shanghai Jiao Tong University), Xiangyu Zhao (City University of Hong Kong)

In the increasingly digitalized world, recommender systems play a crucial role in processing, understanding, and leveraging vast amounts of data collected from the Internet. By accurately modeling user interests and intentions based on their behavioral data, recommender systems can substantially improve user experiences, drive user engagement, and ultimately boost revenue.Recently, we have witnessed that deep learning-based approaches have been widely applied to empower recommender systems by better leveraging the massive data. However, the data utilized in recommender systems typically comprises a large volume of users, items, and user-generated tabular data, which is high-dimensional and extremely sparse. This contrasts with dense data processing applications, such as image classification and speech recognition, where deep learning-based approaches have been extensively explored. How to mine, model, and inference from such high-dimensional sparse data becomes an interesting problem. Furthermore, leveraging such data with deep learning techniques could be a new research direction with high practical value. The characteristics of such data pose unique challenges to the adoption of deep learning in these applications, including modeling, training, online serving, etc. As more academic and industry communities have initiated endeavors to address these challenges, this workshop will offer a platform for researchers and engineers to discuss and identify the obstacles, utilize the opportunities, and propose innovative ideas for the practical application of deep learning on high-dimensional sparse data.