Skip to content

Latest commit

 

History

History
54 lines (28 loc) · 1.34 KB

README.md

File metadata and controls

54 lines (28 loc) · 1.34 KB

ImpactGPT

The base code was forked from https://github.com/tomasonjo/NeoGPT-Explorer.

Info on this project can be found here: https://medium.com/neo4j/knowledge-graph-based-chatbot-with-gpt-3-and-neo4j-c4ebbd325ed

Intro

An AI chatbot that takes user input and generates a query to return framework data.

User inputs relating to Proof Points, Schemes or Claims Data will prompt GPT-3 to generate a cypher query.

The database will be queried and results returned.

Queries not relating to the framework will prompt GPT-3 for a response directly and is more unpredictable.

How to Run

Run the following commands to start the services:

  1. Create an .env file and input your OPENAI API KEY as shown in env.example

  2. Start docker services

docker-compose up
  1. Open localhost:8501 in your browser

Example questions

What is the vegetarian proof point?

What is the evidence for the vegan proof point?

My company uses renewable energy, which proof points would i be eligible for?

All our products are recyclable, which proof points might I be eligible fo

Data

The app is connected to our neo4j knowledge graph instance currently containing scheme data.

Preprocessing

The information extraction pipeline was executed with Diffbot API: See notebooks/Preprocess.ipynb.

Feedback

Please create an issue if you have any feedback!