PSNAppz/SmartResponse

Smartreply

Smart Reply A.I using Deep Learning and NLP techniques.

---

Setting up the environment

• Use virtualenv (install using apt-get install virtualenv)
• Install python3.8 (apt-get install python3.8)
• Create virtual environment using virtualenv venv --python=python3.8
• Activate virtualenv using source venv/bin/activate
• use python api.py for testing

Training the A.I Model

• Update the intents.json file with new training data
• Run python training.py
• Model file will be generated and will be saved as smart-reply-data.pkl
• Thats it!

Using the Flask-API

• Run python api.py
• Use cURL or Postman to test the api running in port 80

cURL request

Predict smart response

    curl --location --request POST 'http://127.0.0.1/api/v1/smartreply' \
    --header 'Content-Type: application/json' \
    --data-raw '{
        "sentence":"Very bad restaurant",
        "username":"Jason",
        "email":"support@mail.com",
        "phone":"9999999999",
        "website":"https://example.com"

    }'

Add more training data

    curl --location --request POST 'http://127.0.0.1/api/v1/update' \
    --header 'Content-Type: application/json' \
    --data-raw '{
        "pattern": "st22323",
        "tag": "spam"
    }'

Add more response

    curl --location --request POST 'http://127.0.0.1/api/v1/addresponse' \
    --header 'Content-Type: application/json' \
    --data-raw '{
        "response": "Hello {user}, please specify your concern",
        "tag": "spam"
    }'