注册 / 登录

Building A Performant Hashtag Ecosystem For LinkedIn

分会场:  爆款架构/数据平台/工程实践

 

案例来源 :

案例讲师

Xin Hu

LinkedIn Tech Leader

Xin Hu is currently an architect and tech lead for the Linkedin flagship feed team. At her current role as senior staff engineer, she has lead (or involved) in multiple large cross-functional initiatives such as Linkedin/MSFT integration, feed infrastructure improvement to speed up product iteration, “world without email” which is a LinkedIn growth initiative to allow members without email address to have a seamless experience on LinkedIn. Xin has a bachelor’s degree in computer science from Peking University as well as a master’s degree in computer science from Bloomington University.

扫描二维码分享案例

 

案例简述

 

As we transform LinkedIn to a place where members form communities around the things that matter most them, hashtags are foundational in enabling members to express what they want to talk about and find others who share the same interests. Hence it becomes imperative for us to improve LinkedIn’s Hashtag infrastructure by heavily investing in relevant domains such as personalization and recommendation, search vertical, feed serving stack, content filtering and SPAM/ Low-quality detection.

 

案例目标

 

While hashtags havebeen part of LinkedIn for a long time and searchable for more than one year, wehave no infrastructure in place to treat them as first-class citizen at LinkedIn. Hashtags cannot be uniquely identified as a key, they cannot be followed and they are never explicitly added as metadata of content. In order to enable further product innovation, it is high time for LinkedIn to start building and operating a performant hashtag ecosystem.

 

成功(或教训)要点

 

Architecture Diagram




 

案例ROI分析

 

By building a scalable and performant hashtag ecosystem, LinkedIn members can easily find, follow, create and engage with communities centered around their interests on LinkedIn. Furthermore, this enabled LinkedIn Flagship App’s core strategy around “active community”, which is to enable members to build and interact with an active professional community to advance their career. This feature has been fully ramped to our English speaking members by the end of September 2018 and resulted in solid impact in user engagement and participation on LinkedIn.




 

案例启示

 

During this session, I will deep-dive on a few sub-projects so that audience will learn the key technical strategies that enabled LinkedIn to evolve its Hashtag related infrastructure into a scalable ecosystem:

●      Provide personalized hashtags follow recommendation to our members.

●      Once a hashtags is followed, content tagged with it automatically start flowing into the member’s feed.

●      Improve hashtag discoverability in the Flagship product, along with onboarding flows on how to tag content with hashtags in the post creation flow

●      Motivate members to add hashtags into the content they posted on LinkedIn, by providing hashtags typeahead and autocomplete

●      When clicking on a hashtag, members will land on a hashtag feed that is backed up by performant search queries fetching all the member posts containing the feed

●      Hashtags marked as spam or low-quality will be automatically filtered out from the product experience or our recommendation system