Skip to main content
Version: 0.1.0

Get Started

Let's discover OML Vision in less than 5 minutes.

Getting Started

VSCode Terminology (Optional)

VSCode Terminology

While not required, it is recommended to read the documentation on VSCode terminology as it will be referenced heavily. Read more here

Install OML Vision

VSCode Marketplace (Automatic Installation)

  1. Go to VSCode Marketplace and install the extension from there.

VSIX File (Manual Installation)

  1. Go to latest release of OML Vision
  2. Download the VSIX file
  3. Install the OML Vision extension using the downloaded VSIX file. Helpful guide here

What you'll need

Windows

  1. Install scoop https://scoop.sh/#/
  2. Use scoop to install git, nodejs, nvm, yarn, java17 and visual studio code if not already installed.
scoop bucket add main
scoop bucket add extras
scoop bucket add java
scoop install main/git
scoop install main/nodejs
scoop install main/nvm
scoop install main/yarn
scoop install java/openjdk17
scoop install extras/vscode
  1. The version of VSCode must be at 1.78.2 or newer.
  2. Install the following VSCode extensions if not already installed
code --install-extension vscjava.vscode-java-pack
code --install-extension vscjava.vscode-gradle

macOS

  1. Install homebrew https://brew.sh
  2. Use scoop to install git, nodejs, nvm, yarn and visual studio code if not already installed.
brew install git
brew install nodejs
brew install nvm
brew install yarn
brew install openjdk@17
brew install --cask visual-studio-code
  1. The version of VSCode must be at 1.78.2 or newer.
  2. Install the following VSCode extensions
code --install-extension vscjava.vscode-java-pack
code --install-extension vscjava.vscode-gradle

Load OML Model Data

OML Vision is designed to work with OML (Ontological Modeling Language) models.

Start from Scratch

Tutorials on how to create an OML model from scratch can be found here

No OML Vision viewpoints defined in above tutorial

OML Vision works with viewpoints which will be covered in Layouts found here

Use existing OML model

If you want to see the capabilities of OML Vision without creating your own OML model you may use one of the following OML models:

  1. Click the eye icon in the sidebar which will activate the OML Vision extension

  2. Your screen should look like this Unopened Workspace

  3. Click Clone OML Model and enter in this URL https://github.com/pogi7/kepler16b-example.git

Command Line Interface

You can also run this command from the command line using

git clone https://github.com/pogi7/kepler16b-example.git
  1. Open the kepler16b-example OML model in VSCode

Start OML Vision

  1. Your screen should look like this Entry Point

  2. Click the eye icon in the sidebar which will activate the OML Vision extension

  3. Your screen should look like this Activate Extension

  4. Data must be loaded into the in-memory RDF triplestore for OML Vision to render views

  5. Click Run clean (This will clear the cache i.e.delete the build folder)

Command Line Interface

You can also run this command from the command line using

./gradlew clean

Look in the build.gradle file for a list of commands

  1. Click Run owlLoad (This will load data into the in-memory RDF triplestore)
Command Line Interface

You can also run this command from the command line using

./gradlew owlLoad

Look in the build.gradle file for a list of commands

  1. Click Objectives
  2. Your screen should look like this Objectives Table