Clone the repo

git clone sudo https://github.com/mirfan899/rasa-qa-bot.git
cd rasa-qa-bot

Create virtualenvs

Run the shell script, which will create two virtualenvs. One for Rasa and second for BERT

sudo ./setup.sh

Services for Bert and rasa

Copy systemctl services files to proper location

sudo cp rasa_run.service /etc/systemd/system/rasa.service
sudo cp bert_run.service /etc/systemd/system/bert.service
sudo cp rasa_actions.service /etc/systemd/system/actions.service

Install SpaCy

We need to install SpaCy.

pip3 install spacy

SpaCy model download and link

Install English model

python -m spacy download en_core_web_sm
python -m spacy link en_core_web_sm en

Train the Rasa model

python -m rasa train

Run Rasa server and BERT server

Enable the CORS for any web api. There is a shell script which and service which will run at as deamon process for deployment. If you setting up this repository, you need to fix the paths in systemctl service files i.e. rasa_run.service, rasa_actions.service bert_run.service

sudo systemctl start rasa
sudo systemctl start actions

Run flask app

For testing purpose run the flask app.

export FLASK_APP=app.py && flask run

Run bert server

sudo systemctl start bert

to stop the server

sudo systemctl stop bert

Adding questions to Rasa bot.

If you want to add questions to Rasa NLU Bot use this file data/nlu/qa.md, there is a script which uses the json format for adding the question into model.

python utils/add_more_questions.py

It will look for file with name english_questions_answers.json with format

[
    {
        "q": "Question here",
        "a": "Answer here"
    },
    {
        "q": "Another Question here",
        "a": "Another Answer here"
    }
]

this script will merge two json files containing questions and answers.

python utils/merge_json.py