How does Foursquare use machine learning to make the product better? originally appeared on Quora - the place to gain and share knowledge, empowering people to learn from others and better understand the world.
Machine learning. What a buzzword! But rightly so for our industry now. We all have access to more and more data, and using machine learning means we can make better decisions for our products based not on what we think is right, but what predictive models uncover about is right.
Many parts of the Foursquare business rely on elements of machine learning, and our superpower, Pilgrim, which understands how people move through the world, was created almost entirely from a machine learning perspective. I'll describe.
In our Foursquare City Guide app, we use machine learning in tip ranking. We spent some time manually extracting signals about tips, such as: What is the sentiment in the tip? Does this tip have a photo? How long ago was it written? How many upvotes has it received? etc. We calculated these features and then had tips ranked by an audience so we could understand how each tip is valued by real users. Those tips were ranked, and then we applied that learning to all of our tips. That's why City Guide tips aren't ranked by date or popularity, but by how valuable they are to our users. So tips are spot-on and relevant as often as possible.
We've been playing around with clustering photos in City Guide too, so that all photos of that burrito will appear together, all indoor photos, photos of people, etc. This too requires machine learning. We label a set of photos, feed it through a learning algorithm and build a model. Hopefully you'll see this implemented in an upcoming release.
But the most exciting way we're using machine learning right now is in our Pilgrim technology. Today, every time a person conducts a search in either Foursquare Swarm (to check in) or Foursquare City Guide (to find a coffee shop), our server does a venue search that computes the probability that you're at a certain place. This also happens when a person does not actually conduct a venue search, but our Pilgrim tech detects that you've stopped into a venue anyway (so Foursquare City Guide can serve up a reminder to order the chai latte, or one of our Pilgrim SDK partners can ping you with a discount promo for the store you're visiting). Once we detect a stop, we predict where you are using machine learning, using signals like visit history, time of day, spatial information, wifi data, Bluetooth beacon placement, and more.
Pilgrim is our superpower, as we said last week in our announcement of the SDK. It's a product that's been in the works for years now, and we truly approached it from beginning to end from a machine learning perspective.
This question originally appeared on Quora - the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter, Facebook, and Google+. More questions: