CUI POC Api.ai Bot

Overview

The Conversational User Interface (CUI) proof-of-concept bot integrates with api.ai on the back-end for intent matching, entity extraction, and slot filling in order to fulfill different conversational dialog scenarios in the medical domain.

More Screenshots

Usage

Environment Setup

It's recommended to download Anaconda 2.7 Python environment from: https://www.continuum.io/downloads

Then setup a matplotlib supporting environment (to support bot's plotting capabilities):

conda create -n matplotenv matplotlib
source activate matplotenv

Run the bot locally

Install dependencies:

pip install -r requirements.txt

Run the bot exectuable setting the slack token from https://robotsandpencils.slack.com/services/B152HETM5:

export SLACK_TOKEN=<YOUR SLACK TOKEN>; python ./bot/app.py

Things are looking good if the console prints something like:

Connected <your bot name> to <your slack team> team at https://<your slack team>.slack.com.

If you want change the logging level, prepend export LOG_LEVEL=<your level>; to the python ./bot/app.py command.

Run locally in Docker

docker build -t starter-python-bot .
docker run --rm -it -e SLACK_TOKEN=<YOUR SLACK API TOKEN> starter-python-bot

Run in BeepBoop

Project page: https://beepboophq.com/0_o/my-projects/84ea846d69d042b3970717f50b7115fd

changes pushed to the remote master branch will automatically deploy.

Run Unit Tests

From the root of the project directory:

python -m unittest discover -s tests -p "*_test.py"

Code Organization

The entry point for the bot is app.py which is responsible for collecting environment variables like log level and slack token, and creating/starting a SlackBot instance from slack_bot.py.

The slack_bot.py module implements an interface that is needed to run a multi-team bot using the Beep Boop Resource API client, by implementing an interface that includes start() and stop() methods and a function that spawns new instances of your bot: spawn_bot. It is the main run loop of the bot instance that will listen to a particular Slack team's RTM events, and dispatch them to the event_handler.py module.

If you want to add or change an event that the bot responds (e.g. when the bot is mentioned, when the bot joins a channel, when a user types a message, etc.), you can modify the _handle_by_type method in event_handler.py.

If you want to change the responses, then you can modify the messenger.py class, and make the corresponding invocation in event_handler.py.

The slack_clients.py module provides a facade of two different Slack API clients which can be enriched to access data from Slack that is needed by your Bot:

1. slackclient - Realtime Messaging (RTM) API to Slack via a websocket connection.
2. slacker - Web API to Slack via RESTful methods.