Application of Data Science at Facebook
Data is the new oil — the most popular phrase in the Data Science community right now. And just like oil, there’s hardly anything that Data can do by itself. Just like oil, it’s available to us in a very crude form which is unusable unless we transform it into usable products.
How we use Data to make choices and decisions is up to us, by extracting knowledge out of it.
Let’s look at how one of the most popular websites/apps (and not just in the social media domain) is using Data Science. Let’s look at Facebook.
It’s your birthday (say), and Facebook will most probably show you a little video comprising of your old photos and posts from your profile. A different video is presented upon completing some years of friendship with someone on its platform. That’s DS! Assuming an average person spends around 2 hours on Facebook everyday (I don’t know if that disappoints you or makes you happy), the 2 hours’ worth of clicks, likes, comments, shares and scrolling is stored in Facebook’s databases for analysis. The objective of this analysis — as we all know today — is to understand the user’s behaviour, improve their experience and create a marketing product exclusively for them (or as close as possible).
Benefits of Data Science for Facebook –
This is how Facebook uses Data Science:
1. Textual Analysis
A majority of data available on Facebook is textual, unlike say, Instagram, where it might be in form of images. Now, FB uses an in-house developed tool called DeepText. This tool analysis the text that we share or post on FB and tries to extract meaning out of them.
2. Facial Recognition
I think we all know where this is applied. Facebook has invested well in facial recognition and image processing capabilities. It can now track its users across its platform and even across the entire internet. The data used to do this is provided by its users themselves by sharing images on their profile or by tagging their friends in other images. The tool used here is the popular one called DeepFace (The Deep — character from The Boys — may or may not be behind these names). FB even claims that this tool is better than even humans when it comes to recognizing faces. Finally, something that can recognize me with or without a beard.
3. Advertising
No prizes for guessing this. Facebook has been using Data Science extremely effectively to decide which ads are shown to which user. The sponsored posts that we see so often these ads are examples of these targeted advertising practice. Search for anything over the web, and Facebook will show you ads related to what you searched for. Probably gonna be seeing ads about Data Science now!
Alright, let’s look at some of the real-life examples of Data Science being used by Facebook:
1. Pride Celebration
As the American Supreme Court passed its judgement on marriage equality, Facebook introduced a rainbow reaction. It allowed users to give a rainbow-like hue to their profile pictures. And about 3 million people used this feature. What does it tell Facebook about its users? They know the interests of these 3 million people and what they believe in. They have more data now and they got this data using Data Science.
2. The Flashback
On celebrating its 10th anniversary, Facebook offered its users to view and share their anniversary of being friends with people on Facebook by making a video of photos and posts shared by them. These were the photos and posts that received the greatest number of likes and comments on the feed.
Let’s look at what information about the users does Facebook collect:
1. First Person Activity
It collects and stores the content and other information that we provide directly. Our personal information that we provide during signing up, information about our relatives, relationship status, phone number, our likes, dislikes, etc.
2. Second Person Activity
This is the information that is provided by others about us. When someone tags you in a photo or a status update. Or shares a video that includes you. Or even when a message is sent to you.
3. Connections and Network Information
Facebook knows that groups and communities that we are connected to and our frequency of interaction with them. That’s why we tend to see more updates (textual, photos, videos) from a certain groups and people.
4. Financial Information
If you’ve done any financial transaction on Facebook, it will collect and store this information as well. Perhaps, you’ve purchased a product or made an in-game financial transaction. Or maybe you’re advertising on Facebook. All this information about your credit/debit card, contact details, etc. are stored with FB.
5. Information About Accessibility
For security reasons (or otherwise) Facebook stores information about how you access it. Do you use a phone or a desktop/laptop? Do you use its app or its website? It collects information about all the computers, phones and other devices where you’ve installed or used its various services including their location, the mobile operator name or ISP, browser used, mobile number, IP address, etc.
This is one of the greatest applications of Data Science and Machine Learning and the reason why Facebook is a giant globally. It tells us why Facebook has achieved the stature that it has by providing unbelievable services to its users!
Ref: data-flair.training
