CircularInsightAI

Hackathon Project - AI EarthHack
Team Members: Jason Jia, Maria Besedovskaya, Nuobei Zhang, Xidan Xu

Summary

We built CircularInsightAI, an AI-powered system that innovatively evaluates circular economy ideas using a multi-faceted filter and enhancer approach. It is designed to empower investors in making efficient and informed assessments.

Final Report

Link to Final Report

Repo Directory

An overview of what each folder and file does:

/data

• Consists of original data, splited data, and code to split data. It also has the manually labeled data set which was used as ground truth
• data/json: consists of JSON version of the CSV file

/preprocessing

• Code to convert CSV file into JSON files, which are saved in data/json

/finetuned_model_test

• Consists of LLM fine-tuning test files and code that generated the plots and accuracy rates

/web_interface

• Self-contained folder containing all files needed to run the Streamlit web interface
• To run the web interface locally, cd to the root of this repo and run streamlit run web_interface/CircularInsightAI.py.

/clusters

• Code of k-means method to get categories of ideas

/api_key_test

• Code to test the GPT API key