R'Eats-Ai

As a college student in a brand new town, you may not know where to find the best restaurants. Fear not Riverside students! R'Eats can find any restaurant close to UCR that you are craving.

Comment

Inspiration

I was inspired to create this project because of my friends' constantly insisting that we eat at Taco Bell. How many times can one eat Taco Bell within a week! The main issue that I realized was that this city was completely new to us, and at our hometowns, we had spent nearly 18 years discovering new restaurants of different cuisines. Therefore, I decided to try and solve that problem with a simple hackathon project.

What it does

This web app allows the user to input their current craving in natural language, and it then gives them the best options that full fill the input in proximity to UC Riverside.

How we built it

I first used Yelp API to ingest all restaurants within 10 miles of UCR and their reviews. I processed this into a natural language dataset that I could feed an LLM(gpt3.5 turbo) the dataset in a readable manner. I then created a script through which a user could input their craving in natural language and prompt engineering using my dataset and my own words in order to display the top options in a user-friendly format. Once the script was ready, I used Flask as my backend to create a web app locally with a Bootstrap template for the UI. I finally deployed this web app with Heroku.

Challenges we ran into

The first and longest challenge I ran into was creating the restaurant dataset. It was easy to create a pandas dataframe, but it felt nearly impossible to extract each and every important field. I was eventually able to identify the most important key words for each restaurant and then used various scripts to display it in usable format.

After that, the main challenges were coding errors, such as iterators not working, not being able to connect with an API key, and finally making sure heroku could actually display the application.

Accomplishments that we're proud of

I am proud that I was able to create this app within a day and that people around me have told me that they would actually use it in their daily lives.

What we learned

I learned more than I ever could from a class about data processing, LLM calls, and even web development.

What's next for R'Eats-Ai

Moving forward, I want to greatly advance the UI past just a template and add a filters field. I also want to enhance the restaurant prediction algorithm by: 1. Adding a hard keyword search before prompting the LLM, 2. Further Prompt Engineering with specifically Retrieval Augmented Generation to make the output geared more towards the users' needs, 3. FineTuning the model a dataset that makes sure it gives consistent and accurate outputs.

I want to add a user login database so they can save their past queries.

Afterwards, I want to add a cost calculation algorithm that takes in the average price of the restaurant and then adds an Uber/Gas cost from your location to see the complete expense of going to the restaurant. One can also ask the App to give lower cost options.

Finally, if I am able to acquire the funds, I want to purchase the full Yelp API subscription so that the app would work for anyone wherever they are by taking their geolocation and developing a dataset using that.

Built With

Share this project:

Updates