Let's be honest, five-star reviews simply don't cut it. Oftentimes the rating system is skewed towards extremely positive or negative reviews, and what is 5 stars in one category may not be 5 stars in another. (You don't want to walk into a restaurant with good food but horrible service, do you?) Yelpie is our solution to this problem. Using natural language processing through the IBM Bluemix API as well as Google Maps and the Yelp API, we are able to find the true meaning behind customer reviews, finding an average eater's sentiment of the restaurant. And not only that, we can even discover a person's opinion behind different specific keywords, allowing you to weigh the pros and cons of each restaurant before picking your dinner for the night.
Inspiration
Being self-proclaimed foodies, we take great pride in scouting out a restaurant, finding out everything we can before actually eating. After a particularly harrowing experience at a "four-star" restaurant, we longed for a way to find ratings for restaurants using specific keywords. Luckily, IBM Bluemix allowed us to do just that.
What it does
Yelpie finds a list of restaurants in a radius around your current location and queries Yelp and Bluemix to interpret a sample of reviews for each restaurant, which is then displayed on Google Maps as a series of color-coded pins. Then, if the user wishes to find a more detailed analysis, they can simply click on a pin to find a keyword analysis for the restaurant, letting them find the sentiment behind each keyword.
How we built it
We built Yelpie using the Bluemix API, the Yelp API, the Google Maps API, and lots of love. <3
Challenges we ran into
We ran into countless challenges along the way; for example, the Yelp API is extremely limited in scope for reviews, a choice we had to work around when developing Yelpie. We also hit our API limit for the Bluemix API, causing unprecedented amounts of stress, but luckily our project was soon back up and running. We found the Google Maps API quite challenging at first, but we slowly learned how to use that as well. At one point, we even had to set up a local node.js server to get our code to run!
Accomplishments that we're proud of
We're proud of learning how to use Bluemix and natural language processing to create a working, useful web app. It was a difficult journey, but we're proud of what we made!
What we learned
We learned a lot about natural language processing, how to use Bluemix, much more about Javascript (neither one of us are very well-versed in Javascript), and much more.
What's next for Yelpie
We want to train Yelpie to recognize categories like food, service, and cleanliness through machine learning and the public Yelp Dataset, so the program can give a much more accurate analysis of what those categories are like for each restaurant.
Log in or sign up for Devpost to join the conversation.