This is a mobile (iOS) app that uses free unstructured voice requests to predict arrival times for nearby trains.
For example, to know the predicted arrival of a train at "Expo Park / USC Station," the station name can be deconstructed to more familiar descriptors such as: Exposition, expo, USC, Coliseum, SC, Jefferson St, California Science Museum, Natural History Museum, Rose Garden, and so and so forth.
The goal of this project is to make querying a live data stream into a more natural extension of language processing.
This app extracts two (2) things from a question a) rail line color (route #) and b) stop description (stop #). These bits of information are then sent to the LA Metro Rail API to display arrival time predictions.
Sample question structure: when will the ____ arrive at _____?
i.e. - When will the gold line arrive at Del Mar Station?
It is built using iOS (Swift), the Wit.ai API for natural language processing, and the LA Metro Rail API.
https://wit.ai/
http://developer.metro.net/