AI That Decides What You Should Eat Based On Sentiment
The Problem
Have you ever wanted to go out to eat with some friends, but no one knew what or where they wanted to eat? So have we. Finding a good place that suits everyone’s preferences and dietary restrictions takes way longer than it should, and then when we all sit down, no one knows what to eat — and that sucks!
Culture
Being from Minnesota, where there’s a culture that takes pride in indecisiveness and defaults to the
“I’m good with what everyone else wants”
disposition, can make things overly complicated.
This problem is so ingrained in everyone that I’ve eaten with people who order the “first thing they saw on the menu” simply because they couldn’t decide on any one dish.
Solution
Automate where you should eat and what you should order when you get there, take the indecisiveness out of the equation — this is the problem we’re trying to solve at IQeats.
I’m a co-founder at IQeats, and what we’re building is a restaurant/meal recommender where we use machine learning to find what restaurants and even what meal items best match your preferences, sentiment, and dietary restrictions so that we can find the best option for you.
So when you’re out partying with the squad — we can recommend the most popping bars. Then, when you’re with your parents at brunch the next morning, hungover, we can recommend the lightest food and the drink with the most electrolytes.
Example
Above is an example of results from the app — searching for restaurants near the mission district in SF.
When we started, we thought that people would be interested only in recommendations based on sentiment — that they’d get a good recommendation and that they’d compare with friends, so on top of the recommendations there’d be a social aspect.
After speaking with some users, however, we found that a crucial functionality we were missing was factoring dietary restrictions into the algorithm, as many people with allergies and on diets find it difficult to find menu items to order.
We’re going to fix that.
On top of this, we’re hoping to incorporate a more social element to the app as well. We have ideas on letting people make comments or add reactions to reviews but we’re still playing with ideas; I have no doubt that we’ll come up with something good — with enough coffee of course!
Insight
I think more important than building a product, is that we are architecting an experience for people. This is important as people decide to go to X restaurant instead of Y restaurant based on some quantitative (price, star rating, etc.) criteria, but also some qualitative criteria as well — like, “will this be fun?”
In this spirit, what we’re really trying to build is a tool to answer the question,
“At which restaurant or bar will my friends and I have the best experience?”
Tech
The technologies we’re using are javascript, html, and a python-django framework. We’re using spaCy, a natural language processor, on the backend to match the similarity of the user's sentiment to the overall sentiment of the restaurant — this can occasionally yield suboptimal results as NLP is only in its infancy, but it does pretty well for how we’re applying it.
Us
We’re three engineers and we want to build products to improve the world. If you have any feedback or ideas (or money to invest in us — we’re not funded so far) let us know! We’d be more than happy to discuss anything and everything, and are very open to new ideas.
Contact info on our website — IQeats
