Elea

Inspiration

As avid chefs, our team was wondering how our cooking impacts the environment. After doing some research, we found that food production accounts for over 30% of carbon emissions. Certain foods, such as meats and nuts, have a disproportionately large impact on the environment. We wanted to make this information easily accessible to consumers.

What it does

Elea is a google chrome extension that allows a user to see the environmental toll the production and shipping of the ingredients used in the recipe they are viewing. They can easily view the relative carbon emissions and water usage of their entire recipe and the biggest contributors to each of those values. Ideally this information would provide another resource for users to access to aid in their recipe decision-making process and hopefully guide them to make more sustainable choices.

How we built it

The frontend of Elea is built on JS/CSS/HTML, rendered in a Google Chrome extension. When a user opens the Elea extension on a website, our frontend sends a post request containing the raw HTML to our backend. The backend is a flask server, which scrapes the recipe ingredients from the HTML. We have implemented two methods for this: a heuristic search, which is quick and cheap but not as accurate, and GPT-3, which is more accurate at higher cost. For the purpose of demonstration, we are using the heuristic search. In the future, we hope to provide better integration with GPT-3. After getting the ingredients, we find their carbon and water footprint from a dataset published in Nature Food. We pass this information back to our frontend, which displays the total footprint of the meal, and the foods that contribute the most to it.

Challenges we ran into

It was very important to us that the user experience was integrated and non-interruptive so that our product could have continued, long-lasting effects. We had originally designed an app and an associated website but, because of these reasons, moved to a google chrome extension because of the grand accessibility. Additionally, it was important that our system could process a data from a large pool of websites so we worked through databases until eventually settling on using AI to help us solve out problem. Leveraging the resourcefulness of GPT-3 and it's training we were able to use NLP to assist or parsing algorithms and efficiently provide the user with helpful information.

Accomplishments that we're proud of

Something that we are really proud of is the way that we were able to establish and make rest API calls to our own http:// hosted servers. This was important because it mirrors a real-world solution and industry database practices. We are also excited that we were able to effectively create a flask backend that links to a javascript and HTML frontend because we were able to learn so much and understand the complex interactions between different languages. Beyond this, overall we are most proud of our ability to create something new that we genuinely believe will help others and make a change.

What we learned

Our team learned a lot this hackathon. We all had very limited frontend web development experience, so we honed our skills in HTML/CSS/JS. We pooled our knowledge to improve our git practices and terminal scripting skills, which improved our understanding of computer networks. We learned the basics of how GPT-3 works, and how to use the API as a developer. We learned how to build a REST API with Flask, and how to use HTTP requests. We continuously iterated our design and idea, with an emphasis on user experience. Finally, we learned about the major problems in sustainability, and brainstormed a variety of small solutions that we could actually implement.

What's next for Elea

Elea needs a few touch ups before initial deployment to the Google Play store. From there, we’ll harness the power of AI on the backend, in order to provide the best possible semantic parsing of ingredients. We would also like to partner with climate agencies, in order to extend our water and carbon footprint data to a larger variety of foods. Ideally, we could find or build a dataset of meals with known environmental footprints, so we can benchmark Elea on ground truth data. In order to have a real impact on sustainability, we’ll need to market our idea. We believe this app is especially well suited towards young adults, who are the predominant users of online cooking sites, and are generally more concerned with sustainability. Finally, we will work on scaling our backend, and serving as many users as possible.

Here is the link to our GitHub Repository

Built With

• flask
• google-extensions
• gtp-3
• html
• javascript
• python