User-review sentiment analysis
User-review sentiment comparison
Individual user food preferences
The inspiration of this hack came from the countless hours wasted on weekend nights trying to plan a place to go out to eat or drink. The complexity of the decision increases with the number of friends in the group. With varying preferences and food choices we decided it was high time there was an application that could help any group finally reach a Consensus!
What it does
Consensus is a web application that uses Yelp-Developer API to retrieve various check-ins and reviews by all members in a group. The application then sends these user-reviews to IBM Watson's Tone Analyzer to determine the sentiment of the review. Based on this analysis each member has a profile that keeps track of their cuisine preferences. When the member joins a group his preferences are compared with the preferences of other members and the application tries to reach a consensus as to what will be the most likely places that the specific group of members would enjoy.
How we built it
Consensus was built completely in python on the back-end. The Yelp- Developer API was used to retrieve food preferences of the members. First, python was used to query the Yelp API for information regarding businesses. Next, python was used to query IBM Watson's analysis API with individual user reviews as the input
The results from IBM Watson were parsed, cleaned and then visualized for better clarity
Challenges we ran into
- Yelp API isn't the most well organized. So that took us a while to get a hang of
- IBM Watson has limitations on query lengths and so we ran into some issues because of that.
- Our team had two members both more comfortable with back-end than Front end UI/UX, hence that was a major roadblock. But like true Hackathon Hackers we did our best to figure it out and ended up with a fairly awesome proof of concept website.
Accomplishments that we're proud of
We are proud of our idea and the meaningful insights we could retrieve from social media activity.
What we learned
We got acquainted with Yelp and IBM Watson API. We learned how to make a website wth decent visualizations
What's next for Consensus
We plan to pull in data from other social media websites. Especially Instagram, since posting your favorite food on Instagram in most cases a higher priority than actually enjoying the meal!