NeoGPT-Explorer

Find more information at: https://medium.com/neo4j/knowledge-graph-based-chatbot-with-gpt-3-and-neo4j-c4ebbd325ed

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

You may run into some errors about not being able to authenticate to pull down 3.9-slim; in that case, double-check that your ~/.docker/config.json file does not have a line that reads "credsStore": "desktop"in it. If it does, delete the line. (For more information see this page.)

3. On the first run, you need to seed the database with the following command; note that you may need to first install pip via conda install pip depenending on how you set up your environment:

sh seed_database.sh

If you are using Windows, you can also execute the notebooks/Import.ipynb notebook for initial import

4. Open localhost:8501 in your browser

Example questions

What are the latest news?

What are the latest news about Apple?

What are the latest news about COVID-19?

What are the latest news about education?

Who was mentioned in a positive light lately?

Most mentioned people in the last 100 articles?

What are the latest facts?

What do you know about Andrew Macintosh?

Where does Ian Chubb work?

What are most mentioned people in sports?

Data

Testing dataset consists of 1000 latest articles from Kaggle repository: https://www.kaggle.com/datasets/adityakharosekar2/guardian-news-articles

Preprocessing

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

Feedback

Please create an issue if you have any feedback!