With Slate Desktop installed, you'll probably want to jump in to create your first translation engine. This guide uses the Dutch to English samples installed in the "Documents\Slate Desktop Samples\Corpus Docs" folder.

  1. Double-click the "Slate Desktop" shortcut on the desktop or in the Start Menu. Slate Desktop starts up and you'll see this window. The first time you run Slate Desktop, the "Engines" list will be empty. Note that there are help tooltip icons (small question marks) that you can click to get more information about an feature in the window.

  2. Click the "Build an engine" button and you will see the "Build a translation engine" form.

    In the "Engine Parameters" section, you enter information about the engine you will create.
    1. Build name: In this box, enter any name you like that will help you identify this engine later. As you type, the UI will tell you some characters are invalid. 
    2. Source language: Enter a 2-character source language code with an optional 2-character locale code. Letters can be upper or lower case. You can use either a dash or underscore as a separator. The system will convert all letters to lower case and convert dashes to underscores.
    3. Target language: Enter a 2-character target language code with an optional 2-character locale code. The system applies the same changes to the target language code.

  3. The "Add files for build" section below, is where you import your corpus documents into the Slate Desktop workspace. You must apply three labels to each file. These labels stay attached to all the data in the files that you import. The UI also tells you some characters are invalid as you type.

  4. When all required label fields are filled (and build name, source, target) without invalid characters, the "Add file(s)" button enables. When you click this button, a browse file dialog opens. You can select files that fit the label values you entered. You can repeat the process as necessary to set new label values and select the files that fit the labels. Note that all files you import during this session become the "build" and the "build name" describes this unique combination of files.

    Different vendors and academics refer to these initial files by many inconsistent names. In Slate Desktop, we call these starting files the "Corpsu Documents." Supported corpus documents include: *.tab, *.tmx and *.xlf (note that sdlxliff is not currently supported). This screen shot shows a completed form with the nl-en sample corpus documents:

  5. Once you import and label all the file for this build, simply click the "Train engine" button. Slate Desktop will prompt you to confirm you want to start the process.

  6. The next window shows the progress through 18 stages of creating the engine. This screen shot gives an example at the beginning. Through the progress, read the changing text and you will learn more about the various processing.

  7. I strongly suggest that you create your first translation engine with some of the sample corpus document samples that we install.