Using natural language processing to route around NYC’s subway disruptions

Gobot analyzes MTA’s service advisories in realtime

Aug 10, 2017 · 5 min read
August 2017 New Yorker Cover
Gobot the robot

Gobot applies natural language processing to MTA’s service alerts

Let’s say you’re travelling from DUMBO in Brooklyn to 59th and Lex in Manhattan, but there’s a service change.

  • But the rest of the F line is running
Gobot analysis

Gobot automatically updates the app to reroute you in real time

A to B routing results are adjusted immediately so that that best possible usable route is ranked at the top!

Automatic realtime rerouting in our app

Gobot constantly figures out the realtime State of the City

Planned work? Sudden power outage? Rush hour trash fire? Gobot is always working at analyzing all subway lines in realtime. Even on weekends.

NYC line service advisory
The real state of NYC

So much wow. How does Gobot do it?

Semantics are tough, especially when they’re inconsistent. The MTA often uses multiple (and abbreviated) names for the same station, describes directions with a confusing mix of actual places, directions and phrases (for example, “Southbound” means “Downtown” which can mean “Brooklyn-bound” which can also mean “Bay Ridge-bound”) and… they sometimes make sbelleng miztakes.

Gobot at work

Don’t worry, just Route

All this happens in the background. Most importantly for users, you just request directions. We will give you the best realtime option, based on what is happening at that moment, without you having to read or decipher anything!

If you can make it here… You can make it anywhere.

New York may be the most challenging city in the world when it comes to disruptions and service changes. But we are also conducting the same analysis on train and bus systems around the world.


The ultimate transit app for major cities

Thanks to Yohann Bensadoun, John Pittock, and Gabriela Stripoli.


Written by

The ultimate transport app!


The ultimate transit app for major cities