Uber如何利用ML和线性编程优化推送通知的时间安排

Push notifications are an integral channel for Uber Eats customers to discover new restaurants, valuable promotions, new offerings such as grocery and alcohol, and the perks of becoming a member, among other things. Push notifications are sent from various teams internally such as Marketing, City Operations, and Product. Since marketing push notifications were introduced in March 2020, not only did the list of teams sending notifications grow quickly, but there was also quick growth in volume to billions of notifications per month by the end of 2020.

推送通知是Uber Eats客户发现新餐厅、有价值的促销活动、新产品（如杂货和酒类）以及成为会员的好处等的一个不可或缺的渠道。推送通知由内部各团队发送，如营销、城市运营和产品。自2020年3月推出营销推送通知以来，不仅发送通知的团队名单迅速增长，而且数量也快速增长，到2020年底，每月有数十亿条通知。

We quickly noticed a variety of issues:

我们很快注意到各种问题。

There were core quality issues (i.e., notifications sent after hours, invalid deeplinks, duplicates, invalid promo codes, landing users on closed stores, etc.)
存在核心的质量问题（即在下班后发送通知，无效的深层链接，重复，无效的促销代码，将用户登陆到关闭的商店，等等）。
Notifications were being sent within minutes and hours of each other, many with conflicting messaging
通知在几分钟或几小时内被发送，许多通知的信息是相互冲突的。
Pushes were being sent to users with little to no personalization in terms of what push the user would prefer to receive, at what time, or at what frequency
在向用户发送推送时，几乎没有对用户希望收到的推送内容、时间或频率进行个性化设置。
Our marketing teams had introduced a variety of new methods to manually control conflicting messaging, adding upwards of 15 hours per week per team member and shifting valuable strategy work to more menial orchestration tasks
我们的营销团队引入了各种新方法来手动控制相互冲突的信息，每个团队成员每周增加15个小时以上，并将宝贵的战略工作转移到更琐碎的协调任务上。

At Uber, we strive to provide a best-in-class user experience, and we quickly realized that a comprehensive approach was going to be required to do this for push. We introduced a system we call the Consumer Communication Gateway (CCG): a centralized intelligence layer to manage the quality, ranking, timing, and frequency of push notifications on a user level.

在Uber，我们努力提供一流的用户体验，我们很快意识到，要做到这一点，就需要一个全面的推送方法。我们引入了一个我们称之为消费者沟通网关（CCG）的系统：一个集中的智能层，...