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.
- Get Slack API Token
- Get Wolfram App ID
- Wikipedia API doesn't require API key.
docker build -t bot . && docker run -e SLACK_TOKEN=<token> -e WOLFRAM_APP_ID=<app_id> bot
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.