Response Classifier on some more ambiguous cases
Early stage of dev: Demonstrating the custom implementation of the functionality of the Google Places Contact API
Later stages of dev: Showcasing the more complex date parsing nuances
I am not an individual who enjoys having to call to find out if my favorite restaurant is open on a holiday. And I'm sure I'm not the only person with that question. That's where yOUpen comes in.
What it does
You text yOUpen whether the particular place your interested in is open on a particular day - it doesn't even have to be a holiday! Then, using natural language processing, it determines the organization you're interested in and on what particular day you want to find out their opening hours. It then discovers the phone number for the nearest location of that particular organization, then formulates a question to ask the employees there. It will then call the company, record their response, transcribe it using a Speech-to-Text functionality, then parse it through a classifier I built to determine whether or not the organization will be open on that day.
How I built it
I utilized the Google Cloud Platform for some of the more intense NLP tasks, and created a REST API hosted on the Google App Engine which was capable of parsing locations from the user's query. This API was called from an express server which was handling the Twilio messages.
Challenges I ran into
So due to the fact I'm using a trial version of Twilio, I am unable to call the numbers I discover. Discovering the numbers was also very difficult, since the Google Places API costs money to use, and costs even more for contact data.
Accomplishments that I'm proud of
I actually managed to implement my own, completely free, version of the Google Places API(with very limited functionality). This was also my first time working with Twilio and the Google Cloud Platform, and I'm pleasantly surprised about how easy and powerful both tools were.
What I learned
That computer generated text readings currently sound like spam calls, even if they have good intent.