NFDI Search Engine ChatBot

Demos

NFDI-ChatBot

The current model operates on Retrieval Augmented Generation (RAG) and integrates optional retrieval techniques like TFIDF, SVM, and Ensemble to fetch documents for language model (LLM) such as Vicuna, enhancing its generation capabilities.

How to run

Run this project by simply following the instructions:

git clone https://github.com/semantic-systems/nfdi-search-engine-chatbot.git
cd nfdi-search-engine-chatbot

conda create -n nfdi_search_engine_chatbot python=3.9
conda activate nfdi_search_engine_chatbot
pip install -r requirements.txt

cp .env-example .env

Modify the .env file and ad your keys and variables there. The run the app:

streamlit run app.py

Then, you can now view your streamlit app in your browser.

Contributors Guidelines

1. Clone the repository to your local machine:

 git clone https://github.com/semantic-systems/nfdi-search-engine-chatbot.git
 cd nfdi-search-engine-chatbot

2. Create a virtual environment with python=3.9, activate it, install the required dependencies and install the pre-commit configuration:

conda create -n nfdi_search_engine_chatbot python=3.9
conda activate nfdi_search_engine_chatbot
pip install -r requirements.txt
pre-commit install

3. Create a branch and commit your changes:

git switch -c <name-your-branch>
# do your changes
git add .
git commit -m "your commit msg"
git push

4. Merge request to main for review.

How to deploy

Using Docker

1. Create a .env file similar to .env-example and add the VICUNA_KEY and VICUNA_URL there.
2. Running using Dockerfile

docker build -t nfdisearchchatbot .

docker run -d -p 6000:6000 nfdisearchchatbot

3. Test whatever everything is set up and works http://0.0.0.0:5000/ping

Using docker-compose

1. Create a .env file similar to .env-example and add the VICUNA_KEY and VICUNA_URL there.
2. Run the following command line:

docker-compose up

3. Test whatever everything is set up and works: http://0.0.0.0:6000/ping

Request to deployed app.

Request URL: http://0.0.0.0:6000/chat Request Body:

{
    "question": "You are talking about who?",
    "chat-history": [],
    "search-results": [
        {

        }
    ]
}

Respond:

{
    "question": "You are talking about who?",
    "chat-history": [{"input": "You are talking about who?", "output": "......."}],
    "search-results": [
        {

        }
    ]
}

Citation

@inproceedings{babaei2024scholarly,
  title={Scholarly Question Answering Using Large Language Models in the NFDI4DataScience Gateway},
  author={Babaei Giglou, Hamed and Taffa, Tilahun Abedissa and Abdullah, Rana and Usmanova, Aida and Usbeck, Ricardo and D’Souza, Jennifer and Auer, S{\"o}ren},
  booktitle={International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs},
  pages={3--18},
  year={2024},
  organization={Springer}
}