LinkedIn offers glimpse at how it prioritizes need-to-know content

As LinkedIn continues to plug away at its digital publishing portfolio -- one of the professional network's chief business strategies -- prioritizing and optimizing the right news and content is critical.

With more than 300 million members worldwide and counting (not to mention a goal to attract virtually every professional on the planet), that is no easy task, to say the least.

The Mountain View, Calif.-based company's engineering team offered a glimpse at a number of moving parts that goes into surfacing the most need-to-know bits of digital content to user news feeds while also trimming the fat.

Looking at notifications as a start, LinkedIn engineering manager Byron Ma explained in a blog post on Thursday morning that there are two requirements for these alerts:

First, all posts must pass our spam and low quality filter before having a notification published for them. Second, only connections whom we deem are strong connections will receive these notifications. We determine this by leveraging the connection strength score from the LinkedIn cloud service. Cloud service maintains connection relationships between members.

LinkedIn also developed what it calls a "Feed-Mixer" system, described by engineers as the most important network distribution channel for member-published articles across both mobile and desktop.

The professional social network cited members who have published at least one article to the social network typically have an average of 1,049 1st degree connections (i.e. immediate, known contacts) and 42 followers.

Feed-Mixer is responsible for organizing a myriad of different types of updates, from shared articles to job listings, into an organized, fluid stream intended to encourage members to keep reading and coming back for more.

Inevitably, this being on the Internet, some junk is going to find its way in.

To minimize spam presence (as well as what LinkedIn defined as "low-quality" content), the site employs a client-side library dubbed the Unified Content Filtering (UCF) Service to dole out confidence scores as to how likely a shared item is a veritable job or event posting, or simply spam.

Ma briefly outlined how content is distributed and organized through other channels. For instance, on LinkedIn's daily and weekly Pulse emails, the system uses Hadoop to run through a member's network of connections and followers to prioritize the most recent articles published by since the last email was sent.