Intelligent Data Assistant (IDA)

IDA aims to become an all-in-one natural language based interface for Data Science Workflows. It's core purpose is to ease the task of a Data Scientist by providing a chat interface. At the moment, IDA is capable of handling the following tasks:

• Load and display the files of a csv based dataset
• Display Bar Graph Visualizations
• Display Force-Directed Graph Visualization
• Suggest and execute scikit's implementation of clustering algorithms

Here's a video demonstration of IDA: https://youtu.be/7FQTiFdqyDE

Coming soon:

• Link to the deployed instance
• Directions on deploying locally

Development quick start

IDA currently consists of three main components:

• ida-ws: A Java Spring server that provides the IDA REST-API.
• ida-chatbot: The Angular webclient for IDA.
• librarian: A general purpose toolkit to programmatically work with software libraries.

You will need at least the following tools:

• bash
• git
• Docker >= 18.09
• Java JDK >= 1.8
• Maven >= 3.6
• Node.js >= 10

Docker is not directly supported by Windows 10 Home and Windows <= 8.1. Use Docker Toolbox if you are using an unsupported OS.

IDA is set up via bash-scripts. If you are using Windows you will have to install bash, e.g. by installing cmder, by installing Cygwin or via WSL.

To get everything up and running follow these steps:

1. Setting up Docker:
   1. Start a local registry: docker run -d -p 5000:5000 --name registry registry:2
   2. Create a local single-node Docker swarm: docker swarm init --advertise-addr 127.0.0.1
2. Starting the IDA stack: ./services/deploy-dev.sh.
3. The IDA web interface should now be available at http://127.0.0.1:3080/.
4. The development stack uses ng serve and spring-boot-devtools internally. Updates to the sources (.class-files in case of Spring) should be reflected automatically.
5. To stop and remove the running development stack run docker stack rm ida-stack-dev.

Production setup

IDA does not yet support distributed environments and is therefore not ready for big production use-cases. To run IDA in a single-node production environment you need:

• bash
• git
• Docker >= 18.09

To start the production stack:

1. Setting up Docker:
   1. Start a local registry: docker run -d -p 5000:5000 --name registry registry:2
   2. Create a single-node Docker swarm: docker swarm init --advertise-addr [SERVER_IP] where [SERVER_IP] is the IP on which IDA should be accessible.
2. Starting the IDA stack: ./services/deploy.sh
3. The IDA web interface should now be available at http://[SERVER_IP]/.
4. To stop and remove the running production stack run docker stack rm ida-stack.