Bot using RASA NLU to answer different questions. Using Wolfram Alpha to get answers. Using Wikipedia as failover.
Article in my blog describing how it works.
Run it with Docker
docker build -t bot . && docker run -e SLACK_TOKEN=<token> -e WOLFRAM_APP_ID=<app_id> bot
Run with Python 3
Install dependencies from Dockerfile and run:
SLACK_TOKEN=<token> WOLFRAM_APP_ID=<app_id> python3 bot.py
Create initial intentions - https://rasahq.github.io/rasa-nlu-trainer/. During the Bot start we run training process based on
rasa-data.json intentions. Later we work with messages we can parse, also Bot stores all unparsed messages so we can check them later.