Part of the training dataset
sus

Inspiration

You must have faced this problem at least once in your life. You stand in front of a rubbish bin with a piece of trash in your hand thinking “Is this thing recyclable?". Of course, it is in everyone’s interest to recycle as much as it is possible. Widespread recycling can help society in a great many ways including overcoming the trash pollution problem and keeping plastics out of the oceans. On the other hand, non-reusable materials found in recycling bins could potentially contaminate sought-after items, like paper and cardboard. They could also cause the equipment to jam up or break down, causing costly damage that could easily be omitted. So, during one of those “Is this thing recyclable?” moments, we thought to ourselves “Why not make things simpler?”

That is how we came up with…

Oxcycle!

Your personal recycling assistant!

What it does?

Oxcycle helps you figure out whether an item is recyclable, non-recyclable or a food waste. You can interact with an Oxcycle bot on WhatsApp which will help you answer the question “Is this thing recyclable?".

How we built it

Dialogflow, a conversational AI, lies at the heart of our application enabling us to understand user's messages using NLP. Then, the requests are fed into the decision tree based on Oxford City Council recycling guide. The tree is written in node.js. After getting an answer, the result is conveyed to the user using a WhatsApp chat bot operated by Twilio.

Challenges we ran into

Surprisingly, the Dialogflow’s documentation was pretty scarce (almost non-existent) but we managed to solve this issue by looking through the source code of the platform. Also, getting ML to work accurately on a really small dataset of predefined user prompts was quite a hassle but we got it working to an acceptable level.

Accomplishments that we're proud of

Even without any of us having prior knowledge of Dialogflow, Twilio or node.js, we managed to build a successful running app. This was also our first hackathon, which came its own challenges and doubts. We are proud of ourselves to overcome this and make a working application.

What we learned

Teamwork! None of us were used to working as a team, but after this experience, we have definitely learned how to collaborate, how to divide work among each other, how to respect each other's work and help each other when one is stuck.

What's next for Oxcycle

We would want to get an approval from WhatsApp/Meta team to make a standalone bot operated by Twilio and not just a Sandbox. Also, we would want to make our bot more accurate by expanding the dataset it uses.