How Entertainment and Social Media Giants are Using Machine Learning

Major names in social media didn’t get there by accident. In addition to their excellent products, marketing, and sales strategies, machine learning is a huge part of the backbone that makes many of their processes successful. Facebook, Twitter, among other names you’ve definitely heard of have become the powerhouses they are partly thanks to in-house data scientists equipped with top-notch machine learning skills. So, how do big names in social media use machine learning?

Facebook

Facebook is pretty transparent about how they use machine learning with their platform, offering an entire research site devoted to ML, among other data science projects involving AI, deep learning, etc. If you have a Facebook, you’ve definitely noticed that many of your feeds, ads, and search results seem oddly fine-tuned towards your own interests. The culprit? Machine learning. Facebook has developed computer vision techniques that essentially read images and videos and deliver content that’s directly related to your interests. So if you see a lot of animal videos that keep you distracted at work, blame machine learning.

[Related Article: Facebook is no longer “social” anything: It’s an AI-Platform that needs to be regulated]

Twitter

Similar to Facebook, machine learning is a core component of what makes Twitter successful. Not long ago, Twitter rolled out features for recommended and highlighted tweets – which are chosen from ML. The machine learning team at Twitter, aka Cortex, uses it to help filter out spam and abusive content to create a streamlined, personalized experience. If you’re like me and can’t stand the “in case you missed it” feature, blame machine learning. If you like it, then thank it.

Pinterest

Pinterest gets a lot of content going through the site daily. That’s why in 2015, Pinterest acquired Kosei, a machine learning company that specialized in the commercial applications of machine learning tech (specifically, content discovery and recommendation algorithms). This led to the creation of Pinterest Labs, where their researchers, engineers, and data scientists constantly build new computer vision models that “see” the content of each to filter abusive and misleading content, rank all of the pins according to relevance, and to help curate the best possible content for all users.

Yelp

You know that new restaurant you went to because you loved the pictures people posted on Yelp? Well, you can at least partly thank ML for that. Not long ago, Yelp needed to find a way to categorize and sort the massive amounts of pictures that the site receives daily. They started by differentiating photos of inside versus outside, food versus drink, and of menus. This lead to changing the UI around to reflect these classifications so users can more easily sort through pictures for what they want. Don’t care about the outside as long as the menu’s good? Need a good interior for that perfect selfie? Yep. Machine learning.

Netflix

“One more episode” you tell yourself as it’s already 1:00AM and you have to be at work at 8:30 the next day. That’s the curse of Netflix that many of us already know. It would be so much easier to watch just one episode if only the service would stop recommending us such oddly-relevant shows to watch next. Do I have to even say what you should thank (or curse) for this? Netflix uses subtle threads within the content itself – as opposed to the broader genres – to help make predictions for what the victim (viewer) should watch next. Hence why although some people might not watch many dramas, for example, may indeed be recommended to watch another drama. It’s not because of the genre, rather it’s because you may be interested in the elements within the content of the drama itself.

Thanks, machine learning.

[Related Article: Why the Current Cambridge Analytica / Facebook Scandal Requires us to Rethink Social Media]

Whether you want to thank ML or not, it’s clear that many media and social media powerhouses are using it in ways that directly affect us every day. These sites/services fine-tune their UIs, recommendation systems, and other features to keep us hooked, which can be a good or bad thing, depending on how you look at it. Regardless of the ethos behind it, we can’t argue that it’s incredibly effective if nothing else.

These sites aren’t the only ones that use machine learning, of course. Consider going to ODSC West this October 31 — November 3 and learn how Reddit uses machine learning.