Conversational Search Bot

This is the code repo for a conversational search bot for book and movie recommendations. It enables natural language queries against Elasticsearch.

Read the accompanying blogpost here.

Useful commands

Train model

rasa train

To use Spacy models

pip install spacy
python -m spacy download en_core_web_md
python -m spacy link en_core_web_md en

Run action server locally

rasa run actions

Run action server from PyCharm

• New Python run configuration
• Module name: rasa_sdk
• Parameters: --actions actions
• ENV vars ES_HOST, ES_USERNAME, ES_PASSWORD must be set

Run rasa shell

rasa shell or rasa shell nlu

Run tests

rasa test

Run cross validation

rasa test nlu --nlu data/nlu.yml --cross-validation

Build Docker image

docker build . -t rasa-bot

docker tag rasa-bot:latest rasa-bot:0.0.1

Add rasa helm repo

helm repo add rasa-x https://rasahq.github.io/rasa-x-helm

Install on local Kubernetes

helm install --namespace rasa --values ./values.yaml rasa-demo rasa-x/rasa-x